I have some results but I'm not confident that they are correct. I am simulating an antenna which should generate a very strong electric field, and I want to test it at the frequency of 200MHz (which is in the range provided by the producer).
I have attached a screenshot from COMSOL depicting ration between E and H norm in the plane XY cutting the antenna in half. The whole sphere has 3 m radius and a PML 0.5 m thick.
You can see that E/H looks like waves in this plane, but I would expect it to converge to the value of the impedance of the free space (~377 ohm). Do you have any idea why it is this way? Is it a matter of mesh quality? Or rather PML settings?
Any suggestions appreciated!
Have you tried simulating using some different dimensions of sphere and PML? Also check the resonant frequencies of the sphere.
Thanks for your answer, I've been trying different sphere radii, but the oscillations remain. One of my experiments was the half-wave dipole. The same oscillations exist even when the radius of the sphere is 12 m for 4 m wavelength.
I do not use Comsol and therefore have no opinion on the simulation details per se. But some speculation: it looks rather similar to a 3:rd over tone induced spherical mode of oscillation. Perhaps the discrete numerics can induce a non linearity, causing the sphere as a system to dispay a standing wave pattern at the 3:rd overtone to the fundamental frequency (as in a varactor frequency tripler)?
Also near field antenna effects are considered important closer than 20 wavelengths from the antenna, so maybe trying a much larger sphere could help if it is possible.
Perhaps it could be interesting to try different mesh densities in order to see how the wave patterns are affected.
Thank you. I think that calculating the model with 20-wavelength PML sphere is over my computational power, though. I have a similar experience with the half-wave dipole, which I can calculate with about 3-wavelength PML sphere. I will try a 2D model of the dipole maybe.
Do you guys suggest to check the resonant frequency of the PML sphere itself? Unfortunately, it's not clear for me how to do it. I will try to estimate the wavelength of those oscillations based on the position of peaks and calculate the frequency out of it.
- Badges
- Science topic