HFSS simulation result fidelity

[pudding]

Hi all friends,
I have simulated an antenna with hfss, but I find the results are different when I set different solution frequencies from 2.5GHz to 3GHz. I don't know why. Which results I can believe. I have uploaded the files and results. Any suggestions or advise will be appreciated. Thanks a lot!
Jianwei

 

[xibillian]

wellcome to HFSS

bkadi(at)gyte.edu.tr

 

[cuongtran]

Normally it must be had the same result, but I think in this case you sould choose the results of the f solu greater, more accurate

 

[pudding]

Hi xibilian,
I simulate again with solution frequency 3GHz, 3.1GHz, 3.2GHz, , but the result still change too much. Do you have some example that the results are consistant to solution frequency? I attached my file again. By the way, would anyone like to check my file? I wonder there is some error leading to the inconsistance. Thanks!
pudding

 

[navuho]

I simulate again with solution frequency 3GHz, 3.1GHz, 3.2GHz, , but the result still change too much

I see the following potential problems at your model :

1. You define high-eps substrate and infinitely thin ground. Such a combination leads to very sensitive behavior, depending on actual mesh. Better to draw real ground thickness and do mesh refinment inside a substrate (+20.000 for example)

2. You define very short input coaxial. Potentialy one could expect high order modes reflection and the length is just not enough to attenuate (it's just assumption, but..). So better to increase the probe length.

3. You define quite large airbox. By defaul HFSS during itterations adds more elements to high field area, thus your final airbox meshing is still worse. Better to
add more elements before first itterations (+20.000 more)

4. It seems that your structure is resonant. In this case it's better to itterate near the resonance and do rather narrow frequency sweep. It almost impossible to get precise result of resonant structure doing broadband freq. calculations. So, divide your freq. band into 2 or 3 and than stich the results.

5. Finally, you set interpolation accuracy to 0.5%. May be it's better to lower it or run pure freq. sweep to check the results.

Have fun with HFSS !

 

[pudding]

Hi navuho,
I appreciate your help. Your comments are very precious for me. Yet I still have some questions:
1. you told me to do mesh refinement inside a substrate. Do I also need to refine the ground and metal's mesh inside?
2. I define the airbox according to the rule that the distance from the radiation element to the boundary should be larger than lambda over 4. If after adding more elements to the airbox, the solving time surges, can I reduce the airbox size?
Thanks a lot!
pudding

 

[navuho]

Hi navuho,
1. Do I also need to refine the ground and metal's mesh inside?

There is no fields inside a metal. You do not need to refine mesh inside,
but you can refine surface mesh as well for better accuracy.

2. Can I reduce the airbox size?

There is no exact answer on this question. I would say you may gradually
shrink the airbox and track the results. It shouldn't be any dependance of the results on the airbox size. When you will see it then you are close to a mininum one.

 

[pudding]

navuho: Thank you very much. Good luck!
pudding