fdtd for dielectric resonator antenna

Hi, I'm simulating Dielectric Resonator Antennas with an in house FDTD code. When the DRA has low permittivity, say around 20-30, few thousands time steps are enough to calculate the S11 parameter. When I use higher permittivities, the more time steps I use the more confused S11 plot I get; with confused I mean lots of resonances/minimums. It seems that less time steps give a better response. I was wondering what is the cause of that: round-off error? coarse meshing?
should I introduce losses in the dielectric/air? what do you suggest?

thanks a lot

hello sir
i also want to do simulation of DRA using FDTD but i m unable to do so ....if possible then plz give me ur calculation/ code or any document which help me to derive its equation and calculating its return loss.......................my email is ravi8331@gmail.com..............
i m waiting for ur response.................

The most likely candidate is incorrect ABC settings. I don't know what kind of ABC you are using
but higher permittivity usually means longer running times and that gives many ABCs problems
(even standard PML has late time instabilities).

"not quite converged but no ABC reflections" usually looks better than "converged + ABC reflections"

There are a few different issues that you could run into with an FDTD simulation that could cause problems:

- Have you verified your original simulations by running them for longer? Your lower permittivity simulations may not have actually converged. The "good" results with fewer time steps may simply be due to lack of convergence.

- Did you remember to decrease your cell size as you increased the permittivity? I often see that people apply the Courant limit based on wavelength in free space. When working with high permittivity materials, you must adjust. You should use a cell size of λ/(10*√εr)

- What boundary conditions are you using? In my experience, the most common boundary conditions (in order of increasing effectiveness) are Mur, Liao, Berenger PML, UPML/CPML. As you use less effective boundary conditions, you need to increase the distance between the edge of your structure and the beginning of the boundary. For particularly challenging geometries (i.e. something like an inductive loop), you may need to increase the padding up to λ/4 at your lowest frequency of interest for Liao or λ/10 for PML.

Tiberius8: I agree with all of your comments. But I don't think we will ever find out
what caused the problems in this case. The initial posting is quite old. I answered
without noticing its date.

That's a shame. I didn't notice the post date either. I would be curious to see how this turned out.

please grendhell,,,,,,, is it possible to upload ur code here............i want to check it
please

Added after 6 minutes:


i'm also very much eager to get that code-----------------please give me that.........email ID is sdpt123_maity@yahoo.co.in

please help me

Added after 4 minutes:



please Tiberius8-----i dont know much more about the boundary conditions according ur reply Mur, Liao, Berenger PML, UPML/CPML etc can u help me-----would u mind to suggest some book where i can get that informations or can u send me that materials--------------my email id is---------------sdpt123_maity@yahoo.co.in