modeling question on basic CP square patch..

[groundhog1]

Hello,

I am trying to make a standard almost square patch with probe feed, where the feed is kind of along the diagonal of the patch.

To simulate/measure axial ratio, I put in a very small pickup dipole element with impedance of 377 some distance from the antenna under test (AUT) face. Then I rotate the antenna under test and watch the S21 between the AUT and pickup.

How come when I rotate the AUT, I don't get the same S21 levels? Shouldn't it stay somewhat constant? I get some really unbelievable variations.

If I rotate the DUT 0, 90, 180, etc. I get very little variation. But on a 45 degree increments, I can get 30dB differences.

I've tried moving the feed all around without making a dent in this problem.

What am I doing wrong?

groundhog

 

[groundhog1]

So as not to confuse..
This is all in simulation. There is no hardware at this point.
gh

 

[Higgler]

Your antenna is linear, not circular. If it was good circular you would probably have one or two dB change as you rotate the patch.
I think you're measuring the antenna based on your response.
Patches made the way you describe are very narrow band. You may just be at the wrong frequency, i.e. you are linear at one frequency, circular at another higher freq. and then linear again at even higher frequency (though it would be the other linear probably).
Locating the feed along the diagonal and making the patch just slightly non square for circular polarization can be touchy if you are building the antenna manually. That's why printed circuit or computer driven hardware is made for narrow band patch antennas, they are very sensitive to dimensional tolerance errors. Only 0.010 inch error may ruin your design.
kch