Discussion Forum

Andreas,

the application library example RF Module > Passive Devices > rf_coil should be a good start point for you.

Cheers Edgar

Hi,

There are several example models in the COMSOL Application Libraries that make use of the Lumped Port boundary condition. If you add a Lumped Port node, you can see that its general tag is lport. If you type

@lport

in the search field at the top of the Application Libraries window and then click Search, the list of models below will be filtered to only include models that include at least one Lumped Port node. See the COMSOL Multiphysics Reference Manual (the "Searching the Application Libraries" section) for more information about how you can use search parameters to find models and applications that showcase some specific functionality. Also see this blog post about searching for a specific model or application .

Best regards, Magnus Ringh, COMSOL

Hi,

I now wanted to calculate the S-parameter of the structure, so I think, I have to use freqency mode. If I do so like in the tutorial rf_coil like mentioned by Mr.Kaiser LumpedPort/Uniform/Cable/on with some additional surface (see picture) I had to make to use it the same way, I see that most power is reflected back to the source and hardly anything leaves the other side of the cable (there is a second identical but passive port). The impedances of the ports are matched to the cable impedance. I think the reflections occur, because the boundary condition doesn't fit right. Although I did a lot of tutorials, I can find nothing that fits better. Please can you tell me what concrete port and what concrete setting is best, so I have no reflections at the port?

Best regards Andreas Erhardsberger

I have attached an example file that may help you along. I did not twist my wires as it might be worthwhile to examine the problem without so much complexity. I don't know what version of comsol you are running but I did this in the latest one. What I am doing here is an open/short connection at the end of the wires with a lumped port at the other end. You have to run two simulations (short and open for end connection) and you can get ind and capacitance, and hence, Zo = sqrt (L/C). Field plots can come once you get impedance being as expected.

Simulation was done at 10MHz, no particular reason. I would suggest getting your simulation to work with a freq domain sweep and then try transient. In order to determine impedance in transient will require FFTs so start in freq domain. Much more effort to troublsehoot in transient and since problem is linear freq domain will be easier.

Good morning Mr Brown,

I've just noticed, that I cannot open your file. When I download it I see that it is a zip archieve. I extract it and there are only filetypes I cannot open with COMSOL.

Best Regards Andreas Erhardsberger

Andreas, just rename it to .mph

Sorry Edgar, but what file? I see a list of 29 files.

Rename the .zip file to a .mph file. Actually Comsol model files are zip archives.

Ok, thanks.

Ok, thanks.

If you do not get this to work for you, maybe I can PM the files to you directly.

dennis brown

Hi,

no thanks it works for me now. I had just to change .zip to .mph. I am currently thinking about it.

Andreas Erhardsberger

Thank you. I didn't think of that way to calculate the impedance. I think it's a very efficient way in regard of computation time. I may have not said it corrctly, but I wanted to calculate the Insertion Loss of the cable for a certain lenth. So I need a second port to get a S21 Parameter. I insered it on the other side of the cable and I get I think sensible values of S11=-13dB and S21=-0.23dB, so almost no reflections. This was after I matched the port impedances to the impedance of the cable. I realize that it is very crucial to use Transition Boundary Condition here. If I use Perfect Electric Conductor on this surfaces instead, like I did before, nearly all power is reflected back to port1. Why is this the case?

Thank you. I didn't think of that way to calculate the impedance. I think it's a very efficient way in regard of computation time. I may have not said it corrctly, but I wanted to calculate the Insertion Loss of the cable for a certain lenth. So I need a second port to get a S21 Parameter. I insered it on the other side of the cable and I get I think sensible values of S11=-13dB and S21=-0.23dB, so almost no reflections. This was after I matched the port impedances to the impedance of the cable. I realize that it is very crucial to use Transition Boundary Condition here. If I use Perfect Electric Conductor on this surfaces instead, like I did before, nearly all power is reflected back to port1. Why is this the case?

I don't know if this is going to be a satisfying answer but one guess (and really just a guess) could be numerical stability in the simulator due to having a boundary condition that is lossy. The PEC is lossless and maybe there is just some noise there corrupting the results.