Air box size effect on the results in HFSS
I have what I find quite misleading results under HFSS. The air box size used for far field computations plays quite an important role on the results.
Of course I use a minimum size which is big enougth ( quarter wave around the structure) , but when I try to vary the size of the box ( bigger box of course) , I find significant differences in the far field pattern ( more than 3db difference on the backlobes) , and last but not least , 10 db difference on s11 !
I tried this on 2 different structures:
-the first one is a loop antenna ( one wavelength torus in front of a one wavelength circular reflector) at 1296Mhz.
-The second one is a conical horn on 24Ghz.
Any idea ?
Dom
sounds like you have created an air box and defined its faces as radiation boundary. ansoft does not suggest this for radiating designs (a bit misleading yes).
without knowing the details of what you're working on i would say that i'm not surprised by differences.
a better way to predict the radiation is this:
build the air box like you described at least 1/4 wavelength from nearest radiating edge. then select all faces and pick PML set up wizard. describe the thickness of pml to be 1/8 wavelength, estimate the nearest radiating edge and describe the lowest frequency of operation correctly.
see the ansoft web site for a month in depth guide of using PML.
Hello ghb,
Few days back I saw a demo of one software.
In that software you dont have to draw your own radiation boundry. software automatically detects that.
So is their any criteria for that or it just make box which is lambada by 4 away from radiating edge?
abhishekabs - im not familiar with that software - but i would suggest yes, in general the larger the box typically larger or equal to lambda/4 the greater accuracy you will get in radiation pattern data
hi i usualy create a box having distance of lambda/4 from the radiating area using vaccum as material assigment and then assign it radiating boundry does this lead to wrong results too ?
This is a general rule, however, I have encountered worse results with larger air box size. Therefore, we may need a stability analysis upon the influence of air box size on simulated results.
Hi All,
While I also use the same lambda/4 @ the lowest frequency for my RadBox spacing, I also like to break it into bands to make it more computationally realizable. For instance if I were to simulate across 5GHz-20GHz I may take the 5-10, 10-15, and 15-20GHz bands and use lambda/4 for the smallest frequency in that band (i.e. 5, 10, 15) this would reduce the box size as neccessary and give you a much smaller memory constraint when working at those lower frequencies.
I hope this helps!
~Nick