Leaky Coax Modeling Issue
Altair EmployeeI am attempting to model a short section of leaky coax with a single slit in the center of the outer conductor with FEKO Suite 7.0. The model is attached. We are trying to evaluate the S21 value between two waveguide ports on the ends of the cable segment but the resulting S21 value is essentially zero (about -115 dB), which I believe is incorrect. When I replace the lossless user-defined dielectric medium of 'ePTFE' with free space in the dielectric of the cable, the S21 value becomes more realistic (about 0.995). I believe there is some mistake in my modeling of the dielectric between the inner and outer conductors.
What restrictions are there on modeling dielectric materials, specifically regarding the assignment of dielectric materials to regions and faces? For instance, in modeling metallic objects only the faces are assigned the desired metallic media of the modeled objects while the regions are set to free space; I do not seem to be able to set the faces of the dielectric to the desired dielectric media (while I can set the 'Jacket' and 'Dielectric' regions to the desired dielectric material) and assume the same rule does not apply to the modeling of dielectrics. What are the 'Default', 'Dielectric Boundary', and PEC face media specifications used for in modeling dielectrics?
Answers
-
Hello riiily,
I had a look at your model and the only problems I see are that some faces that should be set to copper has not been set and are currently set to 'Dielectric boundary'. If I set the following faces to copper, I get a realistic answer of S11=-13.277 (0.2168) dB and S21=-0.22 dB (0.9749).
- Face82
- Face83
- Face84
I don't know why you say that you can't set these faces to copper. I'm using the latest version of FEKO (2017.1.2) and I encourage you to do the same, but there should be no difference w.r.t. what faces can or cannot be set in version 7.0 vs. 2017.1.2.
I believe there is some mistake in my modeling of the dielectric between the inner and outer conductors.
No, the dielectric is fine, but the metal faces were not set correctly.
What are the 'Default', 'Dielectric Boundary', and PEC face media specifications used for in modeling dielectrics?
It is important to understand these concepts and I'll explain them below.
'Default' means that the face type will be derived automatically. The default face type in CADFEKO is PEC - this is easily seen by simply creating a cuboid and noticing that the faces are set to 'Default', but the model shows a PEC coboid. CADFEKO shows the region as PEC, but since we know it is using the MoM, the 6 faces will be written out as PEC and the region medium will be the same everywhere (you may want to read up on the equivalence principle for more info on that topic). If you change the region medium of the cuboid to a dielectric, notice that the faces are no longer PEC, but they have become dielectric faces even though the face medium setting is still 'Default'. This is because the face medium has been derived from the region medium. If the 'Default' or derived value is not the one you wanted, you can change it. 'Default' will always evaluate to one of the other concrete face types.
'Dielectric boundary' simply means that the face is a boundary between two dielectric media. This could represent the face between two dielectric cuboids or the face of a cuboid in free space or any other face between two dielectric regions, as long as the face is not PEC or metal.
PEC means Perfect Electric Conducting and thus it is a perfect metal with infinite conductivity. Since it has infinite conductivity, there is no need to specify a thickness, since all currents flow on the surface (zero skin depth). When modelling metals, usually start by assuming that they are PEC and when required, change them to a lossy metal such as copper.
What restrictions are there on modeling dielectric materials, specifically regarding the assignment of dielectric materials to regions and faces?
I have already replied to this in
I don't understand why you think you cannot assign the media as you need to. I have indicated above that you need to assign Face82 to Face84 to copper and then you get reasonable results. I have also suggested that you upgrade to the latest version, but that I don't think the version makes a difference (I don't remember a bug or something that would prevent you from doing the same in FEKO 7.0).
0 -
Thanks JIF,
I am able to set my dielectric faces to copper. I meant to say that I was unable to set them to say, ePTFE or TEFZEL, the dielectric media I set the regions to be whereas I am only able to assign metallic media to faces of objects and not to the regions of objects.
So the rules are, metallic objects are modeled with free space as the region and the desired metallic substance of the object as the face media and dielectric objects are modeled with the dielectric media assigned to the region and a 'dielectric boundary' as the face media if the dielectric shares the face with another dielectric region or free space and as a metallic media if the dielectric shares the face with a metallic object. Will these rules hold true generally?
0 -
Yes, that is correct. The reason why metals need to be set on faces is that the MoM requires a thickness to be specified. In the future we hope to be able to derive the thickness from the model geometry / mesh so that the user does not have to specif it and then we can allow users to set it on regions (as one would expect).
0 -
I've attached a new version of the leaky coax segment. The segment has a waveguide source on one end and is terminated with a 50 Ohm structure on the other end. I keep receiving the following warning: 'WARNING 3397: The power loss is larger than the active power'. The warning makes me question the validity of the results (which are not exactly what I expected numerically).
In the past I have found this warning has occurred with this model with lossless dielectrics and has been fixed by making the loss tangent of the dielectrics a small nonzero number. In this case however, I am using copper as the only lossy medium (as a friend did in a simulation of the same structure with COMSOL). Why are the losses exceeding the power inputs and does this incorrect result have significant physical meaning? Obviously the model has nonphysical results, but does that invalidate the results entirely? What are the unrealistic features of the model causing this incorrect calculation? Can this situation be resolved without making the dielectrics lossy (to maintain equivalence with my friend's COMSOL model)?
How can the solution of the wave equations even result in the power loss being greater than the active power, even with lossless dielectrics?
0 -
This conversation was continued in:
0