SEMCAD, CST, XFDTD, FEKO or IE3D for UWB simulation?

[adel_48]

advantage feko cst

Hello,

I am trying to simulate two vivaldi antennas (one Tx and the other is Rx). I am sending a UWB pulse from one of them and receiving it from the other antenna.

I have been doing this using CST, but it takes somehow long time, specially if the two antennas are placed very far away. So, I was thinking about other possible software to use. I think time domain simulators such as CST or XFDTD should be the easiest and may be faster (for shorter distances). However, MoM based simulators could also be feasible (obtaining S21 and thus the transfer function) specially when the distance between the two antennas becomes very large. I also heard about SEMCAD and that it is faster by 20-80x, but I am not very sure if this is for all cases or not.

Which simulator do you recommend for such a problem ? any suggestions will be very appreciated.

thanks in advance,
Adel

 

[Manjunatha_hv]

feko hardware acceleration

Use Vector Fields Concerto for such applications...
Concerto 3d EM package which supports the following 3D-EM Solvers
such as MoM (Method Of Moment), FEM (Finite Element Method) and FDTD (Finite Difference Time Domain) Solving Methods in a single environment ....

 

[RFSimulator]

semcad

Hi!

CST (time domain solver) , XFDTD and SemCad are all transient solvers using volume mesh techniques. Because you have to mesh the vacuum in-between the antennas a long distance between the antennas will increase the calculation time accordingly.

Using the MoM based solver uses only a surface mesh and there is not need to mesh the volume between the antennas. At some distance the MoM solver will be therefore more effective in this setup.

I guess this is one reason why CST will add a MoM based solver in their next release. MoM Solver can not directly calculate the time signal as TD codes but you using a FFT (as mentioned in your post) allows you to extract the time signals. From this point of view MoM could be more effective if you indeed want to calculate the entire setup.

In CST you alternatively could use another setup. If you use a farfield probe, you can calculate the resulting time signal at a very long distance away for the transmitting antenna. If you then also would like to take into account the receiving antenna, you can simply use a plane wave excitation with the probe time signal. This requires two small simulations but it might save a lot of time.

This might also be possible in other TD solver but you have to check yourself since I only use CST.!

Concerning SemCAD: Were did you get the information from that it is 20 to 80 times faster then any other time domain solver. I don't believe this! Are you referring to the new hardware acceleration they have announced?

Some other companies are using the same accelerator card and since most part of the algorithm is actually hard-coded on the card and the speed should be pretty much the same for the hardware acceleration for all codes. I have seen the press release and it seems that they are referring to a large speed increase compared to they own coded. For all other software vendors the time decrease is not that drastically and since the hardware speed is rather similar, they leads me to the conclusion that the FDTD loop was not particular effective programmed in SemCAD before. Maybe any Semcad user can give us any insight into this?


F.

 

[Manjunatha_hv]

semcad fdtd

Hello Adel,


Here is the requested very useful document (benchmark.pdf ) on vivaldi antennas using CAD tools you have mentioned (HFSS, CST, Vector Fields Concerto, etc)...



---manju---

 

[mwmmboy]

cst mom solver

Why you are trying to simulate both antennas at the same time'
In my opinion you can save time optimizing both the antennas separately. the accuracy of every antenna can be greater, after you have optimized them you can use some calculation or tool about the link to have an idea of the whole system.


Bye

 

[adel_48]

semcad cst

Hello all,

first thank you all for your replies. I really appreciate your time.

Hi Manjunatha_hv,

I have tried quickwave (FDTD solver of Concerto) before, but the software is very non user friendly. Also, for small enough problems, I couldn't find much speed advantage over CST. Though in larger problems, one of my friends found it much more faster. Maybe the MoM solver is better, but the problem in MoM solvers is correct feed modeling. The only MoM solver I know that can emulate FDTD or FEM waveport is FEKO. Does concerto has this feature too ? since using lumped element at the feed causes some capacitance that should be de-embedded.

Hi RFSimulator,
Thank you very much for your idea. It is actually very useful when the antennas are far enough from each other. I have already seen a paper doing such a thing. But I have also found a paper talking about the antenna transfer function using only the field of the transmitting antenna (you need only to simulate one antenna) do you have some idea about this ?

Thank you again and best regards,
Adel

Added after 3 hours 30 minutes:

By the way, does anybody know if IE3D can simulate those two antennas, or it can simulate only one antenna with finite substrate ?

 

[Manjunatha_hv]

xfdtd

Hello Adel,


Yes, Vector Fields Concerto has following input (excitation)

1. Wave Port
2. Lumped Port
3. Incident Wave
4. Voltage / Current and
5. Magnetic Bias for Ferrite Simulations

Have you tried the new version 6.0?

---manju---

 

[adel_48]

semcad vs xfdtd

hello Manjunatha_hv,

yes I know, but this is for the FDTD simulator (quickwave) I have used quickwave ver 6.0. I was just curious if this is the case also for its MoM solver ?

thanks a lot.
Adel

 

[Manjunatha_hv]

fdtd simulation with cst

Hello Adel,

Lumped Ports (point sources attached to conducting wires) are the only type of port is available for CLASP models i.e MOM analysis in Concerto...



---manju---