CST - Time domain and Frequency domain gives different results of Dipole antenna

[VDinventing]

Hello everyone,

I opened a Dipole originally simulated in Time domain (TD) solver in CST's Component library and found that the Frequency domain (FD) gives completely different results.
This difference persists even though I have tried many different setups i.e. mesh, add more space, boundary,...
I tried other antennas in the library like the Circular patch and the 2 solvers give well-matched results.
As far as I understand these 2 solvers should give not too different results. And that's not valid for this Dipole example of CST.
I've been struggling with this problem for so long and couldn't find any clue to solve it. As I don't fully understand it, I'm not confident about any simulation result given by CST.

Please help me! Thank you very much!

 

[FvM]

The advantage of dipole antenna is that impedance and center frequency can be well predicted from theory. Which simulation corresponds to ideal dipole?

 

[VDinventing]

The advantage of dipole antenna is that impedance and center frequency can be well predicted from theory. Which simulation corresponds to ideal dipole?

The one with the Time domain sir, which is the default when I open this example. The red TD curve with center frequency is close to 27GHz. But the freq domain solver always gives shallower and drives away from this point.

 

[MissingPropeller]

How many wavelengths away are the airbox boundaries?

 

[VDinventing]

How many wavelengths away are the airbox boundaries?

Just lambda/4, I left everything as default. You can open this Dipole in CST component library, it comes with Transient solver and all the setting is there.

 

[MissingPropeller]

Just lambda/4, I left everything as default. You can open this Dipole in CST component library, it comes with Transient solver and all the setting is there.

What results do you get if you increase the distance to lambda/2 or lambda?
I primarily use HFSS and FEM solvers and sweeping the airbox larger until the solution converges is a useful step.

 

[VDinventing]

What results do you get if you increase the distance to lambda/2 or lambda? I primarily use HFSS and FEM solvers and sweeping the airbox larger until the solution converges is a useful step.

I did full lambda (1 wavelength) and it's still the same 🙁

 

[MissingPropeller]

Hmmm. Without knowing the technical details about CST's FEM solver I can't confidently tell you why there is a discrepancy.
I think you should rely on the time domain solver for now though.

 

[VDinventing]

Ah I have the solution now 🥹
It turns out that the distance between the dipoles is too large for the Discrete Port length to be taken into account in the frequency domain solver.
So I ended up having to shrink this distance to less than a certain fraction of a wavelength which was quite frustrating since we had to modify the geometry of the original design just to make it work properly for FD.

 

[volker@muehlhaus]

which was quite frustrating since we had to modify the geometry of the original design just to make it work properly for FD.

The influence of port parasitics and physical port size can be rather large, this applies to all EM solvers. You need to study these effects carefully, and can't rely on defaults for these important modelling details!