How good is the CST Microwave Studio performance?

C$T Performance

Hi all,

anyone tried to use C/S/T Microw@ve Studio in relation to a body that is >> larger than wavelength. E.g. Perform antenna simulation on an aircraft? I will like to know if you encountered any frustrations such as insufficient memory or long simulation time like I did?

Cheers,
Element7k

Re: C$T Performance

Element7k said:

Hi all,

anyone tried to use C/S/T Microw@ve Studio in relation to a body that is >> larger than wavelength. E.g. Perform antenna simulation on an aircraft? I will like to know if you encountered any frustrations such as insufficient memory or long simulation time like I did?

Cheers,
Element7k

Click to expand...

Structure ~10x10 lambda was the hardest I ever simulated. It takes 6 millions of cells, 1.5 GB of RAM and weekend to finish :crazyeyes:
In fact, results are in very good agreement with measurements.
I think more suitable is to use GTD for large objects as does FEKO, FDTD has its natural limitations..

Rgz,
Eirp

Re: C$T Performance

Hi Eirp,

1.5G? It must be quite a big structure! 8O What were your mesh settings? I noticed that CST uses a lot more mesh cells if you haven´t got a planar structure (such as a curved body of an aircraft) I had to use staircase mesh for the imported sheet body. Till now, I still haven´t got the hang of disabling automatic mesh and using maunal mesh. Do you have any advise for me on generating manual mesh apart from the example in the advance topic manual?

Referring to the agreement with measurements, what measurments are you actually to?

Cheers,
Element7k

eirp said:

Structure ~10x10 lambda was the hardest I ever simulated. It takes 6 millions of cells, 1.5 GB of RAM and weekend to finish :crazyeyes:
In fact, results are in very good agreement with measurements.
I think more suitable is to use GTD for large objects as does FEKO, FDTD has its natural limitations..

Rgz,
Eirp

Click to expand...

Dear Element, I have to admit that I tried to simulate an elliple roughly 1.5 by 1 meter length to width when exciting with a dipole at 1.5GHz inside the elliptical solid.With the minimum mesh requirements and with 2 magnetic planes I managed to get a quite handy simulation.Forgot to tell you that this was a SAR mesurement simulation and a sprerical head was involded in that simulation..With a P4 2..2 and 1.5Gs of ram it took me somewhere between 2 and 3 days.This simulation never run in HFSS!!!!!

Regards
vagee

Re: C$T Performance

Element7k said:

Hi Eirp,

1.5G? It must be quite a big structure! 8O What were your mesh settings? I noticed that CST uses a lot more mesh cells if you haven´t got a planar structure (such as a curved body of an aircraft) I had to use staircase mesh for the imported sheet body. Till now, I still haven´t got the hang of disabling automatic mesh and using maunal mesh. Do you have any advise for me on generating manual mesh apart from the example in the advance topic manual?

Referring to the agreement with measurements, what measurments are you actually to?

Cheers,
Element7k

eirp said:

Structure ~10x10 lambda was the hardest I ever simulated. It takes 6 millions of cells, 1.5 GB of RAM and weekend to finish :crazyeyes:
In fact, results are in very good agreement with measurements.
I think more suitable is to use GTD for large objects as does FEKO, FDTD has its natural limitations..

Rgz,
Eirp

Click to expand...

Click to expand...

Hello, Element!

Structure was double ridged horn simulated (and also working there) 200MHz-2GHz
Global mesh settings was If I remember 12 cells/lambda and some mesh refinement around coax to H waveguide. I'm satisfied with automesh, I only make fine refinement if needed. I also look visually at the mesh.
We only make comparison with measurements of reflection coef. and we plan also compare far field pattern.
Regards,
Eirp

Re: C$T Performance

I am interested in the radiation pattern simulation vs measured info as well for the ridged horn. Was it waveguide or coax fed? Did you have good input impedance agreement?

I have had trouble in the past HFSS with the run time of medium size structures and correlation of radiation performance for radiating structures (horn, helix, patch).

Re: C$T Performance

H3O said:

I am interested in the radiation pattern simulation vs measured info as well for the ridged horn. Was it waveguide or coax fed? Did you have good input impedance agreement?

I have had trouble in the past HFSS with the run time of medium size structures and correlation of radiation performance for radiating structures (horn, helix, fix).

Click to expand...

Input impedance fits well with computed one.
Radiation I still don't know, I will ask my friend.
It was coaxial fed. (Coaxial - H waveguide - ridged horn).
Best,
Eirp

Re: C$T Performance

If impedance is good, then radiation pattern would agree better.

A tougher case would be a problem combining large scattering body and some small or curved features. Any experience on this?

Re: C$T Performance

According the horn I'm sending far field.. H plane @1GHz

BLUE - measured
DOTTED RED - calculated MW$.

From curves I think distance when measuring hasn't adequate enough to be in real far field.

Best,

Eirp

Re: C$T Performance

The overlay in the forward region looks pretty good. The simulation issues I have had (HF$$) have been with the calculated response in different regions -backlobes, x-pol, radiated axial ratio ect. In some cases I have had better correlation with WIPL although it can be a bear and you can't rely on it for dielectrics. If you have any x-pol overlays (in the principle and 45 degree planes) I would like to see how well C$T calculates the data. It would also be nice to see the variation in calculated response for these parameters between cst and hfss. ------- I have done a fair amount of fooling around with the rad boundaries, sym axis, mesh seeding and pml but still haven't seen the results that the actual data shows. I could send you a model if you don't mind running it.

H3O

Re: C$T Performance

In attachment is VSWR (dotted red - MW$ and the rest DRhorn - 2 slightly modified types). Note - antennas were quite tuned (to extend to lower frequencies) when built so experiment doesn't 100% fit the measurement.
Dear H30, if you have interesting project in HF$$, please export it as .sat and I can import and calculate via MW$.. + send the model in HF$$
Best regards,

Eirp

PS: Antenna has been measured at next few frequencies, I will provide the comparison soon

Re: C$T Performance

The rest of radiation patterns (E/H planes at 1GHz, 500MHz and layout of the antenna)...