IE3D question
for instance i've designed an annular antenna for gps app's say spaced 30 mils from ground plane and could not match ie3d results with actual measurements.
however if i redesign this antenna and space it say 100 mils the simulation results and actual measurements are quite close.
i donot do filter design however i would expect ie3d to have no problem handling high q designs due to zeland touting their filter simulation comparisons
any comments
Firstly try to put mesh by more than 15 cells/wavelenght.
Also make real ground metal plate meshed with your antenna too improve division on edges because edge effects.
Anyway, post here your ie3d file and referent papers.
Hi...
The error may be due to the infinite substrate. change to air may be obtain better result
I think YYKK should be correct. The infinite substrate assumption may cause difference. You may want to check the IE3D 10.1 to be released before the end of the year. It will have finite substrate (full 3D) capability.
hi, really the ie3d 10 have done the finite substrate boundary?
Not yet. It will be on the IE3D 10.1 to be released before the end of the year.
its realy a good news.who can introduce mom with finite substrate?any papers?
It is not from a paper. It is completely derived and implemented in ZELAND internally. We compared the results with the infinite substrate cases and the results agree very well when the substrate is not too small. By the way, I am Jian from ZELAND.
It seems to me the infinite substrate explanation is counterintuitive. Shounldn't we expect the effect of finite substrate to be more significant when the antenna is spaced further away from the ground? But ghb described that results for 100mil spacing are bettern than those of 30mils.
More details of the design might be needed to get a better explanation.
The reason for the effect of finite substrate is the following: In reality, the substrate is always finite. In the regular multi-layer substrate MOM modeling, we assume it is infinitely extended in the horizontal direction. The infinite substrate will create surface waves. When the thickness of substrate is thin, the surface wave is relatively small. When it is thick, the surface wave is becoming significant. For the finite substrate case, the surface wave will hit the boundary of the finite substrate, part of the power will be radiated out. However, part of it will bounce back to the patch. Normally, the bounced back power should be small and it will not affect the input impedance. However, when the thickness is too large or the size of the finite substrate is too small, the bounced power may be big enough to change the behaviour of the input impedance.
On the other hand, ghb (in 1 post of this topic) observed that the agreement between IE3D and measurement is "quite close" for 100mil spacing, and ghb couldn't match the two for 30mil spacing. The "finite substrate" explanation doesn't apply in this case. There must be some other cause.
I thought ghb said that he got good agreement with thin substrate. Sorry. Normally, the infinite substrate assumption works better with thin substrate as long as it is not too thin. IE3D has a special implementation for ultra thin substrates and it should be good for thin substrate. ghb can send me (jian@zeland.com) the file and I can check it for you.
Hi. Anyone care to share how to go about modeling finite substrate in IE3D? Thanks.
Hi, Jian, I have few questions regarding IE3D:
What is the maximum number of unknowns can be supported by IE3D? Since it is based on MoM methods, so special treatment must have been done for solving the generated dense matrix.
And please give some hints about the simulation time and the cost of buying IE3D.
Best Regards,
Hi, Asdfaaa:
There are different options for the IE3D. If you choose the standard one, it will use a very efficient symmetric matrix solver. It can solve about 16000 unknowns using 2 GB RAM. However, for some structures, you can use iterative matrix solvers. We have an 8 by 8 antenna array with about 16000 unknowns. We were able to solve it using about 120 MB RAM. From what I can see, for similar configuration, we should be able to solve more than 100K unknown problem using 2 GB RAM. Best regards,
hi ghp,could u pls send (upload ) ur geometry, i could help u in ie3d
try to simulate with meshed ground not sheet ground,i think u understand that that reduce the time and simulation momoy
regards
0
Hi ...
This errore may be due to the far field performance
of the antenna.