FDTD simulation of cylindrical waveguides and absorption of longitudinal fields

[peponas]

Hi all,

I have developed a FDTD code employing CFS-PML layers for the BC, and simulating waveguide structures. For this reason, the PML conductivity is scaled along z-axis and as a consequence, only the transverse field components (Er,Ephi,Hr,Hphi) "feel" the PML layer. By using a waveguide mode as an excitation (hard or soft source, doesn't matter), the fields are well absorbed the the PMLs, with minimal reflections as expected. The problem rises with the current sources, especially axial current sources. The PML is transparent to axial components so the Ez and Hz are not being absorbed at all, in problems with excitation with axial current distributions. At this point, I have two questions:

1. Is it correct to introduce current sources (of every orientation) in the PML layer?
2. Is it possible to absorb longitudinal field components with such a formulation and PML? I know about charge accumulation at the ends of finite current distributions which produce DC fields, and that's what I'm looking to mitigate-absorb

In my simulations, I extend the current distribution in the PML layers as well, however the goal is to simulate an almost-DC current distribution.

Thanks in advance, do not hesitate to ask me more details!