Jaakko Juntunen

Reduction of numerical dispersion in FDTD method through artificial anisotropy

In this paper, a simple and computationally low-cost modification of the standard finite-difference time-domain (FDTD) algorithm is presented to reduce numerical dispersion in the algorithm. Both two- and three-dimensional cases are considered. It is shown that the maximum error in phase velocity can be reduced by a factor of 2-7, depending on the shape of the FDTD cell. Although the reduction procedure is optimal for only single frequency, numerical examples show that the proposed method can also improve the accuracy significantly in wide-band inhomogeneous problems.

An improved thin-wire model for FDTD

An improved thin-wire model for the finite-difference time-domain method is proposed. The new model can be used to accurately model straight wire sections connected to other metal structures. In addition, the model includes the effect of charge accumulation at wire end caps. The end-cap model is based on conservation of charge and Coulombs law. Using the end-cap model, unconnected wires such as wire antennas are also accurately modeled. The results indicate a significant improvement in predicting the resonance frequency of a dipole antenna.

Relationship between the compact complex and real variable 2-D FDTD methods in arbitrary anisotropic dielectric waveguides

The relationship between the compact complex and real variable 2-D FDTD methods used for the analysis of guided modes of arbitrary anisotropic dielectric waveguides is investigated. Situations for the permittivity tensor with different non-zero elements are discussed. It is found that in certain cases the complex 2-D FDTD method cannot be reduced to the real variable one. This, in turn, reveals that the real variable 2-D FDTD method has limitation when applied to arbitrary anisotropic dielectric waveguides. In addition, numerical results show that using the complex impulse in the excitation is not an essential condition, even for a purely complex 2-D FDTD method.

Zero reflection coefficient in discretized PML

In this paper we present a closed-form reflection coefficient for the perfectly matched layer (PML), when realized using the finite-difference time-domain (FDTD) algorithm. Examining the reflection coefficient, it is found that zero reflection can be obtained for isolated pairs of frequency and angle of incidence.

An accurate 2-D hard-source model for FDTD

This letter describes an accurate two-dimensional (2-D) hard-source model for finite-difference time-domain (FDTD). The proposed model allows accurate control over the effective radius of a 2-D hard source. In addition, the TM version of the source model is directly applicable as a very accurate 2-D thin-wire model. The proposed model is verified in 2-D for both TM and TE case using a recently introduced method for finding the effective radius of a filamentary 2-D hard source.

Coarseness error in FDTD thin-wire models

We investigated coarseness error related to the finite difference time-domain (FDTD) thin-wire models. We showed that due to coarseness error wires appear longer than the nominal length, and that this error can be considerably larger than that caused by numerical dispersion. Finally, we showed that coarseness error primarily depends on the ratio of wire radius and transverse cell size.

Comments on "Generalized material-independent PML absorbers for the FD-TD simulation of electromagnetic waves in arbitrary anisotropic dielectric and magnetic media" [and reply]

For the original paper see ibid., vol. 4, no. 8, p. 268-270 (1994). In the aforementioned paper, it was claimed that the perfectly matched layer (PML) scheme described in the paper is material-independent (the absorber was thus called the MIPML). The commenter thinks that this is, to some extent, a misleading and inaccurate statement. He explains his reasons for adopting this stance. In reply, the authors contend that the comments given by Chen are incorrect and present an analysis to clarify their generalized MIPML.

On the FEM treatment of wedge singularities in waveguide problems

In modern micro- and millimeter wave technology, there exist many different devices based on microstrip lines. These include coplanar waveguides, patch antennas, filters, power dividers, directional couplers etc. In this work, a simple and computationally advantageous extension to a standard 2D polynomial finite element basis is introduced to cope with wedge singularities in uniform cylindrical structures. To enhance the computation, variable-order elements are used.