feko modeling and verification

Hi there,

I have recently found a design for a quadband antenna on a website of at the University of Queensland.

http://innovexpo.itee.uq.edu.au/2002/projects/s4000751/

Simulation was done in FEKO and the graphs showed promising results. However, I think the loading of the antenna with feko LZ card is in error. There was no measured data for verification that the antenna is working.

FEKO code available in appendix is partially wrong but if anyone wish to have the source which I have retyped. I will upload it.

Also, I will like to know if there are any other fellow FEKO users or anyone else who shares my opinion that the simulation is in error.

Cheers,
Element7k

Hi
I try to simulate it by HFSS and Ensemble .but results in thesis were different.

Hi kamilia,

Yes, I think the thesis is in error. Maybe you will like to upload your results so I can compare with the ones I generated by FEKO. If confirmed in error, I think we should write to the co-ordinator of the university :P and request that he make a note just in case someone decides to use bulid the antenna and find that it is not working.

Cheers,
Element7k

Element7k, everything is OK. :D

This example is proof that MOM tool are the best EM tool for antenna today.
Other like FDTD, FEM will be good for 5-10 years with 10 times fastest processors than processors today.

I'm try in CST and results are OK, but solver time...

In your posted CST file, mesh is set on 50, do you have supercomputer?

Really, how time you need for 20 freq. points in posted FEKO file for this antenna(click on "view output file" and than on " total times")

For CST file just do this:
1. set frequency from 2 to 3 Ghz
2 for fast solver disable adaptive mesh and in mesh properties all set on 10.

How somebody can optimize antenna in CST or HFSS whan this sw. are so slow, I don't know.

Let's see benchmark on ANSOFT site.

For cassegrain reflector antenna from : www.ansoft.com/adf_uk/AMS2002_KClarke.pdf

they used Sun Enterprise 4500 (8 Processors, 8 GB RAM) 8O and they need so long time, where is optimization?

I'm try and with IE3D and results are good too, 100 of 1000 freq. points in several seconds, less than .1 sec per freq. piont.

Crucial thing is Adv. Symmetric Matrix Solver (SMSi), and it's available only on PC based platforms, so with ALPHA or some others workstatons forget on this.

Goxy, thks for your reply. However, I think u are referring to the 2.4Ghz bluetooth antenna which was simulated and not the quad band antenna in the Queensland webpage. :)

No doubts I agree that MoM is a good tool but you might like to read the recent published article on EMC and BEM Enginnering Education: Physics-Based Modeling, Hands-on Training, and Challenges by Cynthia Furse in the Apr 2003 issue of the I/E/E/E antennas and propagation magazine. Now, MoM is superiour because of the hybrid methods at high frequency, however, in a recently paper by T.Weiland, he has proposed optics method at high frequecies as well. How well it will perform, I don't know. Has anyone had a sneak preview at C/S/T MWS 5.0?

:( far from that goxy. I am operating only on a 1.0Ghz P4. 50 mesh ratio is a default setting by CST. (In this case) I think it is to differentiate the dielectic from the PEC. Logically to what I know, it will not have a big effect on simulation time if dielectic and PEC are too much different in thickness.

Total times in FEKO <1min.

In CST, I have doubts that it will reduce simulation time if you start frm 2GHz. If you use the transient solver, I think CST uses FIT and leap frog algo so it is actually an FDTD technique . I am right to say starting from 0 will actually reduce simulation time? I think I had adaptive mesh on in my attached file because I wanted to enable the software to generate the best mesh mesh to compare with FEKO. I usually leave it off.

Both C/S/T and H/F/S/S has additional optimising packages which will run and vary parameters overnight if you want to do big simulation runs. Which I think is useful.

I don't know much about Cassergrains. But I do know that to simulate these sort of antenna you need the curve structure to be quite exact to get good result and they are usually huge. Maybe you might like to tell me more about this. I'll interested to know.

Do you know the engine behind the SMSi? I'll like to know why they can't implement this solver in workstations?

Cheers,
Element7k

Yes, you are right I'm looking in Bluetooth antenna not in this Quadband antenna.
I see that Feko file has error so please upload good Feko file.

Before several month I done cassergrain system in feko with phisics optics method, that was combination of horna, concave and convec metal plate.
Phisics optics method is very good tool in situations like this. See example 35c if you are not.

I don't know anything about engine behind SMSi. They are not probably compiled for RISC processors and so can't be used on this platforms.

Here is the feko code to the quad band antenna.

See second shorting strip(P21,P22,P23,P24), it's not good connected with ground plane.
They are loading 50ohm resistor on wrong place too.

Hi Goxy,

Corrected the gnd connect but results still not desireable. :?

I realised loading is incorrect? Where do you suggest the loading to be placed then?

Have you used the AR/AP card in feko? If you have any experience in using that card I have a small request and was wondering if you can post an example here.

Thanks in advance.

Element7k

50 ohm resistor for matching can not be placed on same place like as excitation(segment 3).
BTW.My recommmendation is that never use resistor loading to match impedance, efficiency will decrease and because gain, better use some capacitive way to match antenna.

Hi Goxy,

elektroda went down for a long time so I'm totally lost. But just to side track on this topic. I have 4 FM antennas placed very close together. Each antenna is used for a different application but they are not active all at the same time, is there any trick that you know of to load the unused antenna so they will not interfere with each other? (Ignoring any switching device that can be used)

Cheers,
Element7k

If your application work about 100mhz, it's very difficult to make isolation because near field of antenna if they are very close.
Also, you can not expect good isolation if they are in the same direction, if they are not affect may be small. Try to calculate effect each on other to see what will happen.
Anyway you can make ground walls beatween your antennas and make better isolation, don't forget to calculate effect of this metal plates.

send more information about your antennas