Discussion Forum

Hi Glenston,

a uniform lumped port is a good choice for a cable connection. You must make sure that the conductors you want to excite by the port are perfect electrical conductors or you apply an impedance boundary condition.

Cheers
Edgar

Glenston,

I suggest we keep it on this thread now.

I checked your model and I think the port setup is entirely wrong. I don't see what the function of the cylinder in the center is. However it is not connected to the spiral or anything. It is just a short circuit.

If you want to excite the spiral you must define a port that connects the two arms of the spiral. You could do that by defining a small rectangle on an xz-workplane that connects to the inner ends of the spiral arms and then make it a uniform port.

If you want to include a coaxial feed, you will need to explicitely include it into the model, define the conductors that connect the coaxial line to the spiral arms and set up a coaxial port across the other end of the cable.

I don't know what CST is. Probably COMSOL has a different way of defining ports. You always need two terminals, the current going into your structure must come out somewhere.

And you will probably need to assign PEC or impedance boundary conditions to the conductors that are connected to the port.

You can have a look into the RF coil tutorial model in the model library that is part of COMSOL. It shows a port definition which is more or less exactly what you need.

Cheers
Edgar

Glenston,

well, this yagi-uda antenna was my first test case for RF and I had assumed I needed the points to plot S-parameters. This is not the case but I didn't remove them because removing or adding geometry elements always messes the boundary definitions up.

Good luck with your spiral!
Edgar

Hi Glenston,

Your current setup creates a short circuit at the end of the feedline. I think you must connect one branch of the spiral to the inner conductor of your feedline and the other branch to the outer conductor.
This is still not optimal because you are feeding from an unbalanced coaxial feed into a symmetric structure but it will certainly result in better S11. You will probably see some common mode on the feed and some mismatch because the spiral's impedance is probably not 50 Ohms in most parts of the spectrum.

Cheers
Edgar

Hi Glenston,

Your current setup creates a short circuit at the end of the feedline. I think you must connect one branch of the spiral to the inner conductor of your feedline and the other branch to the outer conductor.
This is still not optimal because you are feeding from an unbalanced coaxial feed into a symmetric structure but it will certainly result in better S11. You will probably see some common mode on the feed and some mismatch because the spiral's impedance is probably not 50 Ohms in most parts of the spectrum.

Cheers
Edgar

Hi Edgar,

Thanks a lot for your reply.
I tried another very simple setup by replicating what was done in the RF Coil model by using a uniform port.
The VSWR values are still coming very high.

It would be great if you could just have a look at it and let me know what has gone wrong.
Is it the meshing?
Ive attached the model in this post.

Thanks a lot for your time and effort.


Regards
Glenston

Glenston,

I noticed a small gap between the inner end of one of the spiral branches and the rectangle connecting it to the port.

You must zoom into the geometry to see it.

Cheers
Edgar

Glenston,

I noticed a small gap between the inner end of one of the spiral branches and the rectangle connecting it to the port.

You must zoom into the geometry to see it.

Cheers
Edgar

Hi Glenston,

Your current setup creates a short circuit at the end of the feedline. I think you must connect one branch of the spiral to the inner conductor of your feedline and the other branch to the outer conductor.
This is still not optimal because you are feeding from an unbalanced coaxial feed into a symmetric structure but it will certainly result in better S11. You will probably see some common mode on the feed and some mismatch because the spiral's impedance is probably not 50 Ohms in most parts of the spectrum.

Cheers
Edgar

Hi Edgar,

Thanks a lot for your reply.
I tried another very simple setup by replicating what was done in the RF Coil model by using a uniform port.
The VSWR values are still coming very high.

It would be great if you could just have a look at it and let me know what has gone wrong.
Is it the meshing?
Ive attached the model in this post.

Thanks a lot for your time and effort.


Regards
Glenston

I have been watching this thread just because I am curious about the end result. With the spiral antenna2 file you posted, the free tet mesh was quite big. I was getting close to 1million elements. I am not very good at doing them but this example gave me a chance to try and use a swept mesh. If what I did works, the amount of unknowns have been reduced a lot. If not, no harm no foul. Maybe you can check this out and see if what I did is ok. If it works, great. You can also go in a change this so that you get more elements. Given that your structure is so thin at 10um, a swept mesh is better than just a free tet.

I do see the little geometry error on the mesh where Edgar was hinting at in his last post. It looks like you geometry is ok but it is odd that you are getting a discrepancy here. Not sure how to fix it if it really is a problem.

I did this is ver4.2. I see that you did it in 4.1. That might be a problem. I can try and fix it tomorrow when I get a chance. Sorry.

Your ver3 file had a couple errors in it. Please see attached file for the corrections.

You needed to use a free triangular mesh and not a free tet mesh. Also one of the mapped distributions boundaries was incorrect. Not sure why your file differed a bit from what I had post with the ver2 antenna but it looks good now. I did this in ver4.2.

I based the swept mesh using comsol's KB at the following location.

www.comsol.com/support/knowledgebase/120/

dennis

Glenston,

I noticed a small gap between the inner end of one of the spiral branches and the rectangle connecting it to the port.

You must zoom into the geometry to see it.

Cheers
Edgar

First off, you can only have one port on at a time. Just turn off either port 1 or port 2. This also means that you can only calculate S11, S21 OR not S22, S12. Since you have a symmetrical structure, you really don't need to calculate it all. You can turn the othe port on if you want to calculate the other way.

You can simplify your geometry by just using two cylinders upfront and difference out the center one. This way you only need to apply IBC's to all of your gold surfaces. Skin depth is pretty small at your analysis freqs. This will also make your model run faster.

The port faces should be meshed fairly fine. I uploaded my mods and hopefully you can open the file. My version of comsol is 4.2.0.228 and what you uploaded was 4.2.0.150. You could also upgrade it with patches if you are on maintenance.

Did you check out this model file on the website? Pretty close to what you want to do.

www.comsol.com/showroom/documentation/model/2192/

Just looking at your file, you will need an airbox around the structure. This is shown in the above file as well. The above is in ver3.5a but it should translate ok to 4.2 (I believe). There is also a step by step pdf file there as well.

Did you check out this model file on the website? Pretty close to what you want to do.

www.comsol.com/showroom/documentation/model/2192/

Just looking at your file, you will need an airbox around the structure. This is shown in the above file as well. The above is in ver3.5a but it should translate ok to 4.2 (I believe). There is also a step by step pdf file there as well.

Hi Dennis,

Thanks a lot for your reply .
The model looks more or less similar to mine.
A question that i have, what does uniform lumped port represent in real life.

Does it represent a TX line input ?
Does it represent a TEM mode transmission?

Regards
Glenston