Characteristic mode analysis of a patch antenna with substrate
Dear all,
I am trying to learn the correct way to simulate Characteristic mode analysis. As many works in literature use FEKO, I am also using it to replicate the results of one of the papers (YANG AND ADAMS: COMPUTING AND VISUALIZING THE INPUT PARAMETERS OF ARBITRARY PLANAR ANTENNAS). Even though they have used their own computational codes, I was hoping that their first example, a simple rectangular patch with a substrate (12 cm × 7.6 cm, h = 5 mm, Air substrate), shown in Fig. 5(a), could be easily simulated.
But when I simulated with the settings suggested in this forum before, the modal results of the CMA solver were not at all close to the results in the paper Fig. 1. I have tried both with finite and infinite substrate/ground sizes (both the CADFEKO files are attached here), but still they are not matching. I am quite sure I am making some mistakes here, so please help me here. Thank you.
Find more posts tagged with
Altair EmployeeHello Saurav,
Not a problem, the issue was discovered to be the frequency sampling (Sampling of 10MHz provided good results) of the model as well as the feed position (x=30mm, y=0mm) of the patch. Attached is a model that provides the correct results. The highest order modes are 1, 2, 3, 133, 149, and 308, seen below:
Additionally, here is a lua script that plots the higher modes that are greater than some threshold.
Respectfully,
Evan Urban
Hi Evan,
Thank you so much for the script and the introduction of the probe. I was also looking into the nearfield-E field, but it seems they are calculated only for the no of modes calculated (10 here), unlike the case of the eigenmodes (all tracked and untracked, 322 here). So, can you please suggest a way to get the E-fields for the important modes (what your script finds: 1, 2, 3, 133, 149, 308). If I just increase the no of modes to calculate, it takes immense time for simulation. Is there any other way? Thanks again.
With regards
Saurav
Hi Evan,
@Evan_Urban I have one more follow-up question. While processing the eigenvalues, I noticed that they are always positive, which is contrary to the expected results. Even the Yang and Adams paper mentions that: 'All the resonant
modes excited by the current filament have negative λm
below resonance frequency and positive λm above resonance
frequency.' Is it because of any simulation mistake I made, or are you getting the same?
With regards
Saurav
Hello Saurav,
Not a problem, the issue was discovered to be the frequency sampling (Sampling of 10MHz provided good results) of the model as well as the feed position (x=30mm, y=0mm) of the patch. Attached is a model that provides the correct results. The highest order modes are 1, 2, 3, 133, 149, and 308, seen below:
Additionally, here is a lua script that plots the higher modes that are greater than some threshold.
Respectfully,
Evan Urban
Hello Saurav,
You should be able to see the E-Field for the modes, I am attaching my full zip file that includes a .psf file that shows the near field request for the modes. Please let me know if you are having any further issues.
Respectfully,
Evan Urban
Here is an image to show the Real and Imaginary component of the first mode eigenvalue:
Update: I got the results by increasing the frequency points and looking into both tracked and untracked modes. As you can see in the attached picture, the modes on the right side are untracked; even if I increase multiple-fold the no of modes to calculate, these modes are still untracked. So, can you please suggest any way to work around this?