How to simulate microstrip antenna in IE3D using RO4003

I'm a beginner of using IE3D and also design a antenna, I would like to ask DR Jian to help me to fix this problem, I fab this circuit in attached file and measure its S11, this is good but I can't get the same results in IE3D simulation. Is there any simulation environment error or special setting such as alignment or mesh...etc. Please give a help and thanks a lot! Is the finite ground setting correct or the probe setting error?

Hi, chiang:

There are a few things you can improve the accuracy or you should try to avoid problem.

1. You didn't choose meshing alignment option propertly for the finite dielectric modeling. It is critical to align the meshing of the finite dielectrics and the metallic polygons for high accuracy.

2. Your negative port is defined on a wide edge. It will cause accuracy problem. I corrected the above 2 problems in the attached square_r1.geo.

3. Your strip thickness and width is quite big (about 1.6 and 3.x mm) for frequency above 6 GHz. The width is quite big compared to the size of the antenna. It does introduce significant ambiguities to the antenna's input impedance. It is not easy to get very consistent results in both simulation and measurement. In the other word, slight change in the feed or some other thing may cause the Zin quite different. You should try to reduce thickness and width for such high frequency.

Regards.

Thanks Dr Jian, I will try to see your correction first and thanks for your assistance.

Added after 3 minutes:

Dear Dr Jian,
Actually I would like to design a new multi band antenna base on this structure, but now I can't do the thing "what I simulate, what I get in measurement" ....

Hi, chiang:

IE3D normally shouild be able to predict the real thing quite precisely. However, you need to match your model to the reality as much as possible. For example. infinite substrate model is normally very good for regular patch antennas because it does match the reality very well. For your Yagi antenna, if the substrate is not very small, the infinite substrate model still matches the near field condition very well. However, it does not match the far field very well because the major radiation is close to theta = 90 degrees. At this direction, the material in simulation is the substrate while the reality is air. That is why you can't get the pattern very well for this direction. You do need to use finite substrate model for it. Regards.

this may help u!