Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Lumped Port S-parameters > 1 w/circuit interface @ port

Please login with a confirmed email address before reporting spam

Good day,

This is a problem I have been dealing with for a while.

I have a fairly basic setup involving 2 coupled antenna. I have extracted my desired S-parameters from my system in question, and have determined optimal impedance values that should be presented at the ports to optimize S21. Now, I want to simulate my coupled antennae, as I expect the field distribution to change under optimal loading (it does).

The problem arises after simulating the coupled and optimally matched system. Now, my extracted S-parameters are > 1 (in all regards, sometimes abs() >1, sometimes real()>1, also imag()>1). I believe this might just be an artifact of the means by which S11 is calculated for a Lumped Port (S11 = (Vport - Vin) / Vport) when a negative/imaginary value is passed by Vin. (Vin is what is presented to the port, Vport is the calculated by line integral value).

Thus far, I can produce unrealistic S-parameters by either (1) presenting a lumped port with Zo = optimal impedance (though, I kind of expect this to produce odd results; now the characteristic impedance of the line is imaginary) or by (2) using a circuit at the input, with appropriate LC matching networks in place.

If anyone has any insight/tricks to get COMSOL to present me with accurate S-parameters, I would appreciate it. Otherwise, I can always use the un-matched ones and fix them in post.

Thank you,
Jason

3 Replies Last Post 18.11.2016, 11:34 GMT-5
Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13.03.2014, 13:08 GMT-4
I have seen unphysical S11 values when I have either insufficiently-fine meshes or highly-reflective conditions (i.e., where nearly all the input power is reflected back). Severe reflections can arise when default conditions (usually perfect conductor) were supposed to have been replaced by absorbing conditions, but actually remained as conductors. That is a common problem. So, re-check all your various boundary conditions and mesh sizes carefully. Also, if you have enough memory, mesh the problem in more detail, especially at the input port, and also consider using higher order elements than you are currently using, if you have enough memory for that. Check what is happening at your PMLs and/or scattering boundaries to determine if they are properly absorbing the power that is incident upon them.
I have seen unphysical S11 values when I have either insufficiently-fine meshes or highly-reflective conditions (i.e., where nearly all the input power is reflected back). Severe reflections can arise when default conditions (usually perfect conductor) were supposed to have been replaced by absorbing conditions, but actually remained as conductors. That is a common problem. So, re-check all your various boundary conditions and mesh sizes carefully. Also, if you have enough memory, mesh the problem in more detail, especially at the input port, and also consider using higher order elements than you are currently using, if you have enough memory for that. Check what is happening at your PMLs and/or scattering boundaries to determine if they are properly absorbing the power that is incident upon them.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17.03.2014, 17:01 GMT-4
The mesh is definitely not the issue. My structure has two ports, one at each end of a circular waveguide (or just think of it as a cavity resonator. Either way...). The radial waveguide is meshed pretty finely and the ports themselves have an edge-mesh defined such that they are even more densely meshed than the waveguide. Also, it's not a reflections issue. The system and S-parameters all behave fine, and comply very well with theory.

The issue is when I attempt to model an external impedance on the port. Essentially, I'm attempting to apply an impedance match. After performing some post processing, I find my optimal impedances. At first, I applied this by modifying the LumpedPort Impedance (Z_0) to a complex figure (I believe this is the root of the problem). Now, my fields behave exactly as I expect them to (field distributions align with theoretical), but my S-parameters are greater than 1.

Since applying a complex port-impedance is likely messing with the solver, I decided to use a circuit instead, using lumped element components in a circuit model connected through the port through a boundary condition. Again, the same result. The fields behave exactly as expected, but the S-parameters are unrealistic. I thought that a circuit condition could help me apply a matching network without affecting the port parameters, but I was wrong.

I am pretty certain the root of the problem is applying a complex impedance to a Port condition. I was hoping to to see if there was a workaround (likely, post-processing to 'back convert' the S-parameters) and will be looking into this.

Thank you for your time.
The mesh is definitely not the issue. My structure has two ports, one at each end of a circular waveguide (or just think of it as a cavity resonator. Either way...). The radial waveguide is meshed pretty finely and the ports themselves have an edge-mesh defined such that they are even more densely meshed than the waveguide. Also, it's not a reflections issue. The system and S-parameters all behave fine, and comply very well with theory. The issue is when I attempt to model an external impedance on the port. Essentially, I'm attempting to apply an impedance match. After performing some post processing, I find my optimal impedances. At first, I applied this by modifying the LumpedPort Impedance (Z_0) to a complex figure (I believe this is the root of the problem). Now, my fields behave exactly as I expect them to (field distributions align with theoretical), but my S-parameters are greater than 1. Since applying a complex port-impedance is likely messing with the solver, I decided to use a circuit instead, using lumped element components in a circuit model connected through the port through a boundary condition. Again, the same result. The fields behave exactly as expected, but the S-parameters are unrealistic. I thought that a circuit condition could help me apply a matching network without affecting the port parameters, but I was wrong. I am pretty certain the root of the problem is applying a complex impedance to a Port condition. I was hoping to to see if there was a workaround (likely, post-processing to 'back convert' the S-parameters) and will be looking into this. Thank you for your time.

Please login with a confirmed email address before reporting spam

Posted: 8 years ago 18.11.2016, 11:34 GMT-5
Hi Robert,

More than two years after your post, I am facing the exact same challenge as you were: I want to simulate a system with a port/lumped port for which the impedance is matched to my system's impedance, but there doesn't seem to be a way to do it systematically on Comsol. Have you been able to find a solution to this since then?

Thanks,
Ruben
Hi Robert, More than two years after your post, I am facing the exact same challenge as you were: I want to simulate a system with a port/lumped port for which the impedance is matched to my system's impedance, but there doesn't seem to be a way to do it systematically on Comsol. Have you been able to find a solution to this since then? Thanks, Ruben

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.