RL-GO modeling of reflector antenna
__Hi everyone. I have difficulties with modeling of dual reflector antenna of big size (radii of main reflector is more than 200 lambda) using RL-GO. Since PO with 2 (or more reflections) is very slow, i have to use RL-GO. The problem is that I can not reach convergence of directivity value in main beam direction. Probably i will address to you with this problem next time.
__Here is the question concerning a model of single reflector antenna solved with RL-GO at frequency 9 GHz. After solvinig this problem, i will return to dual reflector one. Geometry parameter's are R=6000.0, f_par=6000.0.
__From FEKO warning in .out file
"...Angular increment to fulfil the convergence criterion (min. number of rays per linear wavelength on the geometry >= 10.00):
Eff. uniform angular increment theta-axis: 0.05004 degrees (uniform grid allows max.: 0.02299 degrees)
phi-axis: 0.05000 degrees (uniform grid allows max.: 0.02570 degrees)..."
i see, that FEKO suggest's values of angular increment's of source's far field pattern pattern less than dθmax=0.02299 degrees, dϕmax=0.02570 degrees. I estimated, that there are about 22.14 point per dθmax (ϕmax=const)and 12.38 point per dϕmax (θmax=const) on the rim of paraboloid. See image i attached.
__Ok i choosed
dϕ =0.015 deg < dϕmax
and different values of angular step
dθ= 0.015, 0.011, 0.010, 0.009, 0.008, 0.007, 0.006 < dθmax.
1) Main question is: why don't i see convergence of results, when decrease angular step dθ. I expected results to converge with decreasing dθ.
Additional questions:
2) From description in help i made a conclusion, that your RL-GO solver look's like physical optics.
3) Am i correct, that your PO solver does not support curvelinear mesh?
Remark:
__I want to point out, that my feeder horn pattern was calculated at frequency 10.6 GHz. Project with reflector antenna is modelled at frequency 9 GHz and FEKO doesn't give me any warnings that this 2 frequencies are different.
__I attached .cfx file of reflector antenna and zipped file with horn far field pattern (forum refuced to attach ".ffe" file).
Answers
-
0
-
With Feko 2022.0.1 there is perfect agreement between the different angular increment settings. Also there is a warning about the wrong frequency of the .ffe file.
0 -
Torben Voigt_20420 said:
With Feko 2022.0.1 there is perfect agreement between the different angular increment settings. Also there is a warning about the wrong frequency of the .ffe file.
__Hello mr Voigt. Yes, i have 2018.2 version of FEKO.
__Have you changed something in settings, except for variable "Delta_Theta_RL_GO", which defines angular increment. Or may be some default settings are different between version's. May be you can suggest some setting to avoid this convergence problem.
__Or, finally, may be there was some fixes after version 2018.2, which solved this problem?
__I attach a table, from which you can see, that results begin to decrease after Number of huygens sources exceeds ≈159*10^6
0 -
Hi @Marat Taram,
I used a mesh size of 1000mm which is absolutely sufficient to represent the geometry. Your mesh is extremely fine but this won't affect the results. I used default RL-GO settings abgesehen von "Fixed grid increments" für Delta_Theta_RL_GO. Please note that also with the recommended "Adaptive ray launching" the same results are achieved. "Fixed grid increments" is actually only necessary if there are very small, important geometry parts or surfaces are very oblique with respect to the incident direction of the rays. In that case Feko will give your mentioned warning. But this is not the case in this model.
There have been several improvements since Feko 2018, hard to say, which one was significant here.
Best regards,
Torben
0 -
Torben Voigt_20420 said:
Hi @Marat Taram,
I used a mesh size of 1000mm which is absolutely sufficient to represent the geometry. Your mesh is extremely fine but this won't affect the results. I used default RL-GO settings abgesehen von "Fixed grid increments" für Delta_Theta_RL_GO. Please note that also with the recommended "Adaptive ray launching" the same results are achieved. "Fixed grid increments" is actually only necessary if there are very small, important geometry parts or surfaces are very oblique with respect to the incident direction of the rays. In that case Feko will give your mentioned warning. But this is not the case in this model.
There have been several improvements since Feko 2018, hard to say, which one was significant here.
Best regards,
Torben
__Ok. Mister Voigt. Thanks for your answer and time. Can you also answer about PO
1) if PO solver support curvelinear mesh? FEKO gives me warning
"Warning 17581: Curvilinear meshing will not be applied to MoM/MLFMM faces while global basis function control is disabled. Global basis function control can be enabled using the solver settings dialog. The "Windscreen" and "Ray launching - geometrical optics" solution methods are exceptions; they do not require global basis function control for curvilinear meshing".
2) "Large element (PO) - only illuminated from front" what does it mean from front. Am i correct, that front is direction to which normal vector is pointing?
0 -
Hi @Marat Taram,
Curvi-linear mesh can only not be used with PO.
Faces have a front and a backside. You can visualize the normal direction of faces in CADFEKO in
Display settings -> Colour -> Element normal
For PO / LE-PO you can choose
- full ray-tracing: the solver checks via ray tracing which PO faces are illuminated by the H-field and which are not.
- always illuminated: the check is avoided and all PO faces are illuminated (even if they are in a shadow region!). Recommended e.g. for reflectors.
- only illuminated from front: the check is performed, but accelerated by considering only the front side of the faces.
To be on the safe side, you can always use full ray-tracing.
Please also check the UserGuide chapter "Physical Optics (PO) and Large Element Physical Optics (LE-PO)".
0
Altair Employee