Microstrip Patch antenna has improved impedance matching but no improvement on gain

[mckeee6]

I am trying to design and understand a 30mmx45mm patch antenna. The setup is below, but it is a 50mm long, 50ohm T.L. feeding the patch antenna on Rogers5870 circuit board. I am trying to do impedance matching with an open circuit stub. My issue is two-fold: i) when I have a really good impedance match, where my S11 parameter is very low, my gain does not increase and ii) the 3dpolar plot is not symmetric. The details are below and I have attached the simulation file

Case 1: No tuning stub--just patch antenna and transmission feed line
The electric field plot and the 3d polar gain plot showing asymmetry, here the gain is 5.8dB

S11 parameter at 3.125GHz is -5dB. I deembeded the excitation by 50mm to the entrance of the patch antenna to estimate what the impedance was and Re(z)=89ohms and Im(z)=-75ohms

Case 2 included the tuning stub used for impedance matching
As before, the electric field plot and the 3d polar gain plot showing asymmetry, here the gain is 5.8dB

S11 parameter at 3.125GHz is now -37dB. I deembeded the excitation by 50mm to the entrance of the patch antenna to estimate what the impedance was and Re(z)=52ohms and Im(z)=-.6ohms which means the impedance matching is working quite well. My question is, why is my gain not improving?

 

[biff44]

a better return loss does NOT necessarily mean the antenna is radiating more power!

Consider if you used a 50 ohm chip resistor as an antenna. It would have a VERY GOOD return loss, but, would not radiate at all--making a very poor antenna.

In your simulation, you should sweep the frequency and look at BOTH S11, and S21 (to some sort of test antenna simulated at some distance). only it you are having a high S21 magnitude, while having a low S11 magnitude, are you truly "tuning" your antenna.

the same is true if you were measuring your antenna patch on a network analyzer in the near field. you can not just tune the S11, you also need to see if the transmission to a target antenna is getting any power as you tune the patch

 

[hagster]

Normally EM software treats mismatch and gain as separate things. You need to plot 'realized gain' that includes the mismatch.

The asymmetry is caused by the feed line. The left hand edge(slot) of the antenna is shorter than the right hand one. The effect isn't huge, less than 1dB in your case. You can get rid of it by having a pin feed.

 

[vfone]

This is true, the "realized gain" includes the mismatch loss, but this is not the situation in this case, because return loss is very good.
Giving the chip antenna as an example it is auspicious, but be careful, they have a lot of fans around the world

Solving this issue, I would start from the statement that (usually) the patch antenna don't need any impedance matching network at all, and for 50 ohms is very possible to design the antenna in the way getting good input impedance. Try using the U-shape match:

https://article.sciencepublishinggroup.com/journal/132/1321009/image001.gif

Rogers Duroid 5870 is a very low loss mmWave material, and due to its relative high cost is rarely used at 3GHz. Better would be one of RO4000 materials:
https://rogerscorp.com/advanced-connectivity-solutions/ro4000-series-laminates