which be better MOM or FDTD

[cybcad]

FDTD and MOM,which is better to analyze the microstrip antenna? i want to know which one is more efficient and more accurate to use in conjunction with an optimization procedure?



thanks

 

[bluuice]

I think FDTD is more easy to anlysis,and can give enough efficient accurate,MOM is more difficient.but also more efficient.

 

[rautio]

Hi bluuice -- I work for Sonnet, we sell a MoM planar tool and we also represent and support CST in North America. CST uses Finite Integration Method, similar to FDTD but without some of the stability problems of FDTD.

Basically, if you can get both, do so, they are very complimentary. Sonnet is a frequency-by-frequency analysis, and CST is time domain. Time domain means that you hit it with an impulse and look at the response and do an FFT. You have to get enough of the impulse response and then you have very accurate broad band data.

If you have any part of your circuit that is not 3-D planar (3-D planar includes vias, but almost all dielectric is layered), then you want CST. You especially want to use Sonnet for a planar circuit if there are a lot of ports (analysis time for most volume meshers increases linearly with the number of ports; for most MoM codes, the number of ports has no effect).

For broad band data, Sonnet has a very good interpolation. For example, a 20 resonator filter can be analyzed at 4 or 5 frequencies and the entire rest of the filter response is interpolated very accurately. Narrow band filters can be a problem for most time domain tools due to the long ringing in the impulse response, however, CST (and a few other tools too) can spot the long ringing portion of the impulse response and extrapolate it out so narrow band filters are not a problem.

Using any volume meshing tool (CST, HFSS, etc.) on a pure 3-D planar circuit is actually seriously ridiculous, asusuming you have access to any kind of planar tool. For example, it is very easy to draw a planar circuit on a 2-D computer screen. Much more difficult to draw a planar circuit in 3-D on a 2-D computer screen. However, if you have, say, a coax connector that is an important part of the circuit, then you probably want to use the volume meshing tool, like CST. Analysis time required to achieve a given level of error for a planar circuit are typically about 10X to 1000X faster for a planar tool as compared to a volume mesher.

There is a wide spread mis-conception that volume meshing tools are more accurate than planar tools for planar circuits with thick metal. This is completely, wildly, totally, and absolutely wrong. The situation is completely opposite of that. Can say more about that if there is interest.

As for analysis error, you can generally get a good idea how good any tool is by looking at the current distribution. It must have very high current on all sharp edges and be smooth. If you need very high accuracy, you should make sure that the current distribution looks good. In time domain tools, just keep calculating the impulse response until the current looks good. In HFSS like tools, just keep refining the mesh until the current distribution looks good.

CST also has a frequency domain analysis that uses the usual time domain style rectagular mesh, and a frequency domain tetrahedral mesh analysis. This wide variety is unique among EM software vendors.

Lots more to say, but this post is already too long. If you want to play with the Sonnet 3-D planar MoM tool, get the free, no time-out version at www.sonnetsoftware.com.

 

[cybcad]

hi rautio
if i use MoM to analyze microstrip antenna, what model shall i take to simulate the probe feed accurately , is VED a good chioce?



thanks

 

[aze]

Lots more to say, but this post is already too long. If you want to play with the Sonnet 3-D planar MoM tool, get the free, no time-out version at www.sonnetsoftware.com.

Hi rautio,

Is it afully functional trial version?
why by following the link above I get this window..., is it a broken link?

 

[kleinesmurf69]

just leave out the dot at the end, thus www.sonnetsoftware.com

greetz

 

[rautio]

Hi cybcad -- If antenna feed is a coax center conductor from below, use Sonnet's vias. If microstrip feed, use a microstrip feed. There is a chapter on antenna analysis in the manual (complete pdf file included with the free download) and fully worked example files come along with SonnetLite (click on help). kleinesmurf69 listed the correct link (thanks!).

Sonnet is an analysis in a box. The top cover can be set to 377 Ohms per square and be sure the top cover is about 1/2 wavelength or more above the antenna. The sidewalls must remain conducting (so the FFT can be used), so move the sidewalls out so the sidewalls do not affect the antenna. This usually requies at least 1/2 wavelength from the edge of the antenna.

Dipoles are tough because they radiate sideways (unlike patch antennas), so the sidewalls have to be moved out quite some distance. Fortunately, all this only increases the size of the FFT, so often there is almost no impact on speed.

You can also use unshielded type analyses (Momentum and IE3D), they use an unshielded EM calculation for coupling (Green's function). While they have no sidewalls to worry about, they do assume infinite substrate size. This is not a problem unless there are surface waves.

If all you have is a single patch antenna, SonnetLite is plenty, the software is good forever (no time out), problem size is just right. You can also get a demo version of IE3D (last I checked anyway) but it has a time out.

Full disclosure, I work for Sonnet. IE3D and Momentum are kind of competitors, depending on how you look at it.

 

[Animesh]

Hi

Both MoM and FDTD methods have advantages and disadvantages.
I had worked on microstrip slot antenna and I would suggest to go for MoM for microstrip antennas.
As MoM are used for simulation of antennas or arbitrarily shaped metallic structures. MoM also provides good isolations if dimentions are very small.

If we compair MoM technique, the FDTD method is better suited for solving those
problems geometries that contain inhomogeneous bodies.

In some places MoM gives problems so the best way is to used Hybrid MoM-FDTD method.

Thanks & Best Regards
Animesh