What is the best EM simulator for MoM?
I am a new member at this forum,i hope we have good discussions.
i have a question:What is the best EM simulator for MoM?FEKO,MiniNEC or something else?My initial geometry is an inhomogeneous dielectric plate.
Than for your time.
EMmaniac
Looks like you don't many options for the problem. Besides Feko, wipl-d might work (www.wipl-d.com).
For start, I think that Sonnet can help, and it is downdable for free and in this configuration it is excellent for smaller geometries and if 2.5 D simulation can do. (www.sonnetusa.com). WIPL-D is good solution for MoM tools, but I don't think that the MoM is best for dielectric structures, even for sccatering.
I am a f.e.k.0 user and I agree with Debeli. For structures with dielectric, MoM not as straight forward as other general techniques like FI or FEM. Lots of intermediate steps before you can get results and one cannot be sure if the intermediate step, e.g. coating is done correctly! However, it has the advantage that it can produce very quick results in high frequency range using optics. MoM best for structures made mostly of metal or wires.
Cheers,
Element7k
Hi to loucy,Debeli and Element7k,thanx for your response.As a researcher. i am interested in FMM which 'comes' from MoM.It is not very hard obviously to use a software package-as i see, CST is good one- and solve the problem.But i am intersted in the basic algorithm firstly and secondly in the application.Anyway,so what is your opinion for the software part?CST, HFSS, or something else?
yours sincerely :o
gaudi
P.S Element7k,did you find the FEKO LITE version from internet or you use the full-version?Is there any way to have it too?
it is difficult to say which one is the best, In my opinion "CST-studio" is very good.
Hi gaudi,
to use a software package is not very hard but to use it correctly and get meaningful results might sometimes be difficult. If u r interested in MoM codes then looking into C$T might be a wrong direction. Anyway they are using FI and keep their algorithms well away from publications. (Anyone disagree on this point?) F/E/K/0 and HF$$ on the other hand does publish their methods occasionally and might even help you out (technically) if you are interested and have proper maintenace license.
Cheers,
Element7k
Hello fellows!
Another powerfull software using conformal FDTD is (not as well known as CST) QuickWave3D (www.qwed.com.pl). Its author claims that he developed CFDTD algorithm even before cst guys...
Their company isn't as strong as cst so its gui is horrible (few ppl must develop all the software alone) but engine seems powerful.
For example, they are using S-par definitions for even lossy lines and cst not (in m*w*s v4.3). Anyway I think coming m*w*s v5 will improve this..
This is just to remind this cst "competitor"..
Best,
Eirp
let's continue the thread on MoM, and particularly FMM (fast multipole method). I have two questions:
Some listed FMM along with FFT etc as one of most important algorithms found in the last century. Do you agree?
Is FMM only applicable to metallic structure, not useful for layered dielectric?
"CST studio" is a excellent tool to computation EM problem.
Hi to all friends,
all your notes are very useful for me.My opinion is that FEKO and HFSS are very good for scattering problems and CST for antennas etc.To come to FMM as loucy said,my research interets at this time is the development of FMM of scattering from dielectric bodies.I agree with you,loucy,that FMM is one of the greatest algorithms found in last century.
Yours sincerely
gaudi :o
Ok, let's talk about the weather
FMM, nowadays, isn't well suited to solve planar multilayer structures, even mixing it with DCIM.
It's very usefull for problems using free space Green's function because you can expand it using the addition theorem using spherical bessel/hankel functions and legendre polynomials. By upgrading FMM to a multilevel scheme, MLFMA, you can reduce the computational complexity of your N unknowns problem to O(N LogN) that is you can compute the induced current due to your helix antenna placed on your daddy's B-52 This is, more or less, the same efficiency you can get using CGM-FFT but without the rigid uniform sampling grid . Some people say you can use DCIM with MLFMA and I have to belive them This is a good field to research in (go and get your Ph D)
Another topic, very very hard, is to use Lindell's EIT (free space green's function and exact image theory for multilayer structures) and FMM or MLFMA. Buffff....
Good luck
Have anyone established an exact image theory for multilayered cases?
To me even the theory for single interface case hasn't seen a lot of practical application. Any further comments?
As far as DCIM (discrete complex images) is concerned, I've read that it has problem for large distance if one proceeds without the surface waves. How do you bring in the FMM if that is the case?
EIT and relatives
As far as I know, there is no such extension for the multilayer case. Only microstrip and half-space problems has been solved using EIT. There is a paper for the multilayer case using EIT but for the quasi-static aproximation (P. D. Einziger et al, "Rigourous image-series expansions....", IEEE-AP, vol 50, No 12, pp 1813-1823, dec 2002) It's a very interesting field; hard to work, plenty of math, oh god.... (We shouldn't forget microstrip structures were first analized using quasi-static aproximations)
Is this useful? Well, it depends on your Ph D director. I think that EIT extended to the multilayer case can be useful to understand why DCIM works That is, why can you say that the function in the kernel of the Sommerfeld integral can be expresed as a series of complex distance placed image currents? Why GPOF, why not Pisarenko? By now, there is no answer for these questions. At least you can accept arguments like "...because you can use Sommerfeld identity..."
DCIM and large distances
I'm not a DCIM guru, but I know a bit about Sommerfeld integrals and I can asure you cannot forget the surface wave contribution from the poles; near, far or wherever you go. In the first Aksun articles, he extracted the poles contribution and after that he expanded the kernel. Nowadays, people are forgetting the poles ???? Why? Because you have to solve a nonlinear equation (zeros of the generalized TE and TM reflexion coefficients denominantors)
and compute a residue (using a numerical quadrature, or analitically) I think it's not the way to go (to forget the poles)
Mixing DCIM with FMM/MLFMA
Let the gurus talk
W. C. Chew et al, "Fast and efficient algorithms in computational electromagnetics", Ed Artech House, 2001.
And that's enough (I think )
why not try other simulation tools . I know Micro-strips which use transmission line matrix method is a good choice.you said "My initial geometry is an inhomogeneous dielectric plate. " I think this software is suitable for you, it is a time domain method hence inherently wideband .
Dear nanjingchenbing
i do not know the software you describe.Could you tell me any site where i i can see some information about it?
Yours sincerely
gaudi :o
He means Micro-Stripes from Flomerics
http://www.microstripes.com/
Hi gaudi
just like chinasky said, the website is http://www.microstripes.com/, you can appreciate it :)
Regards
nanjingchenbing