Discussion Forum

Posted: 1 decade ago 13 giu 2011, 09:22 GMT-4

[QUOTE] Hi, I am trying to study the behavior of a semiconductor resistance (having a resistance of 38.5 ohms at dc) at THz frequencies for which i extracted the S parameters using the port boundary condition. I would now like to convert these S parameters to Z parameters for which i would like to know what is the load and a source impedance that comsol is assuming while calculating S parameters ? Is it 50 ohms as some other softwares assume? I read in the RF reference guide that it assumes that all ports are assumed to be matched and that there are no reflections at the ports directly. So does that mean that it is 38.5 ohms in my case ? Is there a option to configure this impedance? would also like to know if comsol can covert these S parameters to Z parameters ? Would be extremely happy if someone helps !! Thank You. Rekha [/QUOTE] For any conversions between different types of two port networks, you will have to do this by hand or enter the formulas in comsol. There is no built in conversion feature as of ver 4.2. I doubt very highly that your Zo is 38.5 ohms at RF. Your other question is a bit more complicated to answer. I believe comsol does default to 50 ohms but you have to have the ports/geometry set up correctly. If you are using regular ports, and depending on the geometry of the structure, you may have to manually calculate this. Example: for a round coaxial line, what sets the Zo of the line is knowing the center and dielectric diameters (could also use radii). There is a standard formula for this in most EM textbooks. Zo = (eta_o/(2*pi*sqrt(e_r))) * ln(D/d) (this assumes u_r is = 1) For round coaxial implementation, you will need to set up an annular ring that has a certain radius, over the center conductor in your geometry. Say your geometry has a piece of wire that you want to connect to, in one approach you could setup a workplane for face parallel to the wire's flat side end (which will have a certain diameter), you will add another circle on top of this with the correct sized radius. You will then have a specific Zo. Also note that this particular situation requires the outer circle to be touching a return path in the geometry. If none exists, you may have to create an airbox. I have left out some details here so you have more work to do. I do not know the case for other versions that are not coaxial lines. If this is for waveguides or something more oddball, I do not have specific examples but the above may be a guide to an approach. Others may be able to help. Suffice it to say that setting this up, while not difficult, is not completely obvious. You may also get more help if you post a file.

Posted: 1 decade ago 21 giu 2011, 11:23 GMT-4

Thank you so much for the response, Dennis !! My geometry is just a solid cuboid and not a cylinder, so i am afraid that the formulas you suggested may not apply here... I am attaching my model file along with this message for a reference. Moreover , i am taking the Impedance to be 50 ohms as you said for my Z-parameter calculations.

Attachments:

• semi_resis_RF_module_2D_1.mph

Posted: 1 decade ago 21 giu 2011, 16:27 GMT-4

[QUOTE] Thank you so much for the response, Dennis !! My geometry is just a solid cuboid and not a cylinder, so i am afraid that the formulas you suggested may not apply here... I am attaching my model file along with this message for a reference. Moreover , i am taking the Impedance to be 50 ohms as you said for my Z-parameter calculations. [/QUOTE] I don't think my formulas would work for your situation. I looked at the file but I am not super familiar with the details of v3.5 as I have been weened on ver4.0 and higher. Importing ver3.5 to higher versions does not always work well in that all things may not translate. In this case, it looks like everything came through. Comsol does not really define what the Zo of coaxial ports are and you have to do it yourself. They don't do that in the newer versions as well. I don't know how to define the Zo in your case. Since your problem is 2D, maybe what you can do is first start with some knowns and work back around to what you want to solve for. If you made your material equivalent to a 1ohm resistor, you can determine s-parameters for that from a textbook from Pozar and then convert to z-parameters. Then follow that for your problem. I generated S11 and S21 below in 4.0a just for reference. 10e5 -0.999999999+1.2499352984e-6i 1.0231745002e-9+1.2493330751e-6i Wish I could be more helpful here. I would like to know how to define Zo for your problem. Once you have that, there are formulas that allow you to convert from s to z parameters.