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.
Waveguide at low frequencies
Posted Jan 26, 2015, 8:15 a.m. EST 7 Replies
Please login with a confirmed email address before reporting spam
Hi!
I am trying to simulate a waveguide and I have some questions. My waveguide has an inner and outer radius of 0.00164 meter and 0.00546 meter, respectively. The medium inside the waveguide is air and the borders are perfect electric conductors. See attached picture. I use the physics "Electromagnetic Waves, Frequency Domain (emw)", and the study "Mode Analysis" and simulate the Electric Field Norm (V/m) at three different frequencies, 1 kHz, 1 MHz and 1 GHz (see attached picture). I also calculate the "Propagation constant (rad/m)" and "Effective Mode Index" at these frequencies.
As you can see from the attached picture the results are not good at lower frequencies. I have three questions:
1. Why is the result so bad at lower frequencies? Is it because of limitations with Comsol? I have tried a finer mesh but that did not help.
2. If Comsol is not the reason then what is? What can I change in order to calculate my waveguide modes, electric field norm and propagation constant in a correct way also at lower frequencies?
3. I have built a 3D-model of a waveguide and I would like to be able to calculate the propagation constant. But, is this possible also for a 3D-model? The "Mode Analysis" does not seem to be available in that case.
I use Comsol version 4.4.
If someone has the interest and possibility to answer my questions I would be very grateful!
Thanks in advance!
/ Stefan
I am trying to simulate a waveguide and I have some questions. My waveguide has an inner and outer radius of 0.00164 meter and 0.00546 meter, respectively. The medium inside the waveguide is air and the borders are perfect electric conductors. See attached picture. I use the physics "Electromagnetic Waves, Frequency Domain (emw)", and the study "Mode Analysis" and simulate the Electric Field Norm (V/m) at three different frequencies, 1 kHz, 1 MHz and 1 GHz (see attached picture). I also calculate the "Propagation constant (rad/m)" and "Effective Mode Index" at these frequencies.
As you can see from the attached picture the results are not good at lower frequencies. I have three questions:
1. Why is the result so bad at lower frequencies? Is it because of limitations with Comsol? I have tried a finer mesh but that did not help.
2. If Comsol is not the reason then what is? What can I change in order to calculate my waveguide modes, electric field norm and propagation constant in a correct way also at lower frequencies?
3. I have built a 3D-model of a waveguide and I would like to be able to calculate the propagation constant. But, is this possible also for a 3D-model? The "Mode Analysis" does not seem to be available in that case.
I use Comsol version 4.4.
If someone has the interest and possibility to answer my questions I would be very grateful!
Thanks in advance!
/ Stefan
Attachments:
7 Replies Last Post Feb 4, 2015, 9:25 a.m. EST
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Hi,
the frequencies are far below the cut off frequencies at those waveguide dimensions even at 1 GHz. So you can't see modes. You would have to go somewhere above 25 - 50 GHz to see modes at your dimensions. Actually your geometry looks like a coaxial cable and you normally wouldn't like to see modes in a typical application.
Cheers
Edgar
--
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
the frequencies are far below the cut off frequencies at those waveguide dimensions even at 1 GHz. So you can't see modes. You would have to go somewhere above 25 - 50 GHz to see modes at your dimensions. Actually your geometry looks like a coaxial cable and you normally wouldn't like to see modes in a typical application.
Cheers
Edgar
--
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
--
--.
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Hi Edgar!
Thanks for your reply!
My type of waveguide support the TEM mode which has no cutoff frequency. The TEM mode propagate and exist also at very low frequencies.
It is true, my model is a coaxial cable, but still also a waveguide. I use here an easy example in order to explain my problem. My actual problem concerns the modelling of a high voltage cable, which is a type of coaxial cable. But, if Comsol doesn't calculate simple problems in a correct way, I don't want to continue with a more complicated case.
/ Stefan
Thanks for your reply!
My type of waveguide support the TEM mode which has no cutoff frequency. The TEM mode propagate and exist also at very low frequencies.
It is true, my model is a coaxial cable, but still also a waveguide. I use here an easy example in order to explain my problem. My actual problem concerns the modelling of a high voltage cable, which is a type of coaxial cable. But, if Comsol doesn't calculate simple problems in a correct way, I don't want to continue with a more complicated case.
/ Stefan
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
The emw-physics is an overkill here. The AC/DC physics should be sufficient for the low frequencies and probably even for 1 GHz.
--
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Hi Edgar!
Thanks for your reply!
My type of waveguide support the TEM mode which has no cutoff frequency. The TEM mode propagate and exist also at very low frequencies.
It is true, my model is a coaxial cable, but still also a waveguide. I use here an easy example in order to explain my problem. My actual problem concerns the modelling of a high voltage cable, which is a type of coaxial cable. But, if Comsol doesn't calculate simple problems in a correct way, I don't want to continue with a more complicated case.
/ Stefan
I don't have a direct answer to your questions but some background reading may be good to do here:
First, there is a magazine article here on page 3 that may help a little bit.
www.comsol.com/zmags/multiphysics-simulation-2014
I believe you can download a pdf of the magazine too.
Second, here is a discussion of AC/DC vs EMW usage. It is a very grey area in what to use sometimes:
www.comsol.com/blogs/computational-electromagnetics-modeling-which-module-to-use/
I think you should consider using ACDC. Doing really low frequency analysis in EMW is possible but it does require a unique setup and a bit more skill. It is very difficult to truly to simulate a 1kHz to 1GHz broadband response because as you go higher in freq the mesh has to get tighter in order to resolve enough of your geometry. Now you can adjust your mesh based on your frequency via some parameters so it is possible to do this but it is just a matter of experience using the program. ACDC and EMW are very similar as both solve maxwell's eqns. There are details in either case that must be addressed so it may not just be very simple. And in any case, post a sample file and perhaps somebody can see if you are doing something wrong or gives hints.
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Hi Dennis!
Thank you for the links! The first one directing me to the article didn't help me, but the second link looks interesting. I will read it and see if it can help me!
As both you and Edgar suggest, perhaps I should try the AC/DC module instead, but I am not sure I can find propagation constants using that module, but I will try!
/ Stefan
Thank you for the links! The first one directing me to the article didn't help me, but the second link looks interesting. I will read it and see if it can help me!
As both you and Edgar suggest, perhaps I should try the AC/DC module instead, but I am not sure I can find propagation constants using that module, but I will try!
/ Stefan
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
Hello Stefan,
You are correct in that the Electromagnetic Waves, Frequency Domain interface (emw) would contain the appropriate formulation to use here. The emw interface is indeed appropriate for wave-type phenomena such as waveguides. From your images, it also looks like you are starting from this example:
www.comsol.com/model/finding-the-impedance-of-a-coaxial-cable-12351
And that example is appropriate to start with, but you do need to check the assumptions of that model.
You are using the PEC boundary condition to model the inner and outer conductor. This boundary condition is appropriate if we can assume that the loss in the metal is negligible, and that the skin depth is much smaller than the dimensions of the waveguide. At 1-10GHz this is certainly reasonable, but as you drop in frequency it is no longer true, the skin depth becomes a significant fraction on the inner and out conductor. Once that happens you will want to model the entire waveguide, metal and air. The lower you go in frequency, the more current flows in the conductors. You may want to check if such low frequencies are typical for this sized coax, since you will have significant current flowing along the outside surface of the coax.
You may also want to look at the analytic solutions for a coaxial cable and transmission lines in general, which should be in any EM textbook. In particular, think about the low frequency regime as approaching the static limit.
Best Regards,
Walter
You are correct in that the Electromagnetic Waves, Frequency Domain interface (emw) would contain the appropriate formulation to use here. The emw interface is indeed appropriate for wave-type phenomena such as waveguides. From your images, it also looks like you are starting from this example:
www.comsol.com/model/finding-the-impedance-of-a-coaxial-cable-12351
And that example is appropriate to start with, but you do need to check the assumptions of that model.
You are using the PEC boundary condition to model the inner and outer conductor. This boundary condition is appropriate if we can assume that the loss in the metal is negligible, and that the skin depth is much smaller than the dimensions of the waveguide. At 1-10GHz this is certainly reasonable, but as you drop in frequency it is no longer true, the skin depth becomes a significant fraction on the inner and out conductor. Once that happens you will want to model the entire waveguide, metal and air. The lower you go in frequency, the more current flows in the conductors. You may want to check if such low frequencies are typical for this sized coax, since you will have significant current flowing along the outside surface of the coax.
You may also want to look at the analytic solutions for a coaxial cable and transmission lines in general, which should be in any EM textbook. In particular, think about the low frequency regime as approaching the static limit.
Best Regards,
Walter