how we can explain this! in patch antenna gain

[Abu Maria.]

hi.
in ads momentum , i simulate patch antenna that resonate at 2.45GHz , with (length=21.9 mm , width=36.26mm,height=8mm);

i use (h=8mm) just to see the effect of increasing the height of the substrate , effectiveley i get 2.45GHz!

but

note that ads work with an infinite ground plane , i get (directivity=7.31dB, Gain=3.74dB) , these results are logical for me ,because in the case of a considerable height of the substrate , there are losses wave propagating in the substrate , which causes a decrease in the gain!

and when i add a finite ground plane , normally it was expected a smaller gain than in the first case (with infinite ground plane) , because in this cas (finite ground) we see the occurrence of the back-lobes that are a losses , so normally we get a gain <3.74dB (first cas with infinite ground)

but i get a gain 4. 2 dB !!!

how we can explain this ? NB: i use ground plan with dimension : Lg(ground)=L(patch)+4*h(height)
Wg(ground)=W(patch)+4*h(height)

thank you!

 

[ferdows]

hi dear abu maria
why u didn't check antenna with full wave simulator like HFSS or CST? ADS good software for RF simulation but its momentum is not very good but if it simulated on infinite ground it should be check by CST

 

[Abu Maria.]

thank you ferdows!

I am familiar with ads , and I do not know how to work with HFSS or cst!

please why you say its momentum is not very good) , since I can replace the infinite ground plane by a finite plane!

 

[Shehata]

ADS-Momentum is based on moments method (with some fast algorithms and approximations) which should be fine for these conductor based passives. However, I also recommend u use an FEM solver like hfss. My rule is to use hfss for antennas and ADS for others (such as filters, phase shifters etc..)

- - - Updated - - -

If you find it difficult to work with hfss you can send me the dimensions/material (if not confidential) and I can simulate it for you

 

[volker_muehlhaus]

ADS-Momentum is based on moments method (with some fast algorithms and approximations) which should be fine for these conductor based passives.

Very true. The method of moments solver are full wave solvers.

The HFSS and CST marketing guys like to talk about their tools as "full 3D", but method of moments solvers like Momentum of course solve the full Maxwell's equations in 3D. The only difference is that they are specialized to solve a planar problem, and this antenna is a planar problem.

In theory, there is no advantage of using HFSS or CST for this problem. However, it could be useful to get another data point for comparison with the conflicting Momentum results.

 

[ferdows]

method of moment is a full wave method but ADS and IE3D not a 3D full wave as i know ADS is 2D full wave and IE3D i saw in some paper called 2.5D
so they assume wave propagation at 2 plane so 3D full wave is best for antenna as shehata said ADS best software for passive element
and as i understood abu maria told the antenna has height so its not microstrip 2D structure

 

[volker_muehlhaus]

method of moment is a full wave method but ADS and IE3D not a 3D full wave as i know ADS is 2D full wave and IE3D i saw in some paper called 2.5D

ferdows, this is almost completely wrong. I don't know why you comment on EM tools without knowing what you talk about.

 

[thylacine1975]

Hi guys - I can't decisively weigh in on the comparison between EM simulators, but I keep stumbling across dozens of papers comparing them out there on the 'net whenever I'm looking for something else

Abu Maria - I'm guessing, but perhaps the high potential directivity of your antenna is a clue? If the match to your antenna has improved (i.e. lower S11) with the reduced groundplane (...although this seems unlikely given your generous ground plane dimensions?), you will thereby couple more power into the antenna, and see a gain approaching the theoretical directivity. With antenna efficiency η and directivity D (and neglecting feed losses etc), realized gain ≈ ηD(1-S11²)

 

[volker_muehlhaus]

patch antenna that resonate at 2.45GHz , with (length=21.9 mm , width=36.26mm,height=8mm);

To reproduce the model with other simulators, we need more information:
What is the substrate epsilon and tand?
What feed type did you use, inset or probe feed? Feed location?
What metal definition did you use?

 

[Abu Maria.]

thank you all !

let me tell you that at the beginning I worked with the antenna(ep_r=4.7 , h=1.6 mm ) and I want to see the effect of increasing the height of the substrate on the gain substrate on the gain , so i increase heigth for h=8mm!


so

when i work with infinite ground plane
L(length)=21.9 mm , W(width)=36.26 mm ; h=8mm , ep_r=4.7 , loss=0 , conductor=PEC (perfect conductor)

when i want to work with finite ground plane (i assumed that :size(ground)=size(patch)+4*h))
S11 change , so I have to play on the length to get the right resonance frequency!

L(patch)=23.5mm with the same dimensions (you can change these dimension until you get the right frequency, its yout choice)!

i alimente patch with port interne in ads momentum!

::::::::::::

ferdows you say (... and as i understood abu maria told the antenna has height so its not microstrip 2D structure ) , can you tell me why

why then in momentum, it gives the hand to bring (height) of the substrate?
why then there is also a ground plane, so momentum is valid for these simulations!

 

[flanello]

If you switch from a infinite ground plane to a finite ground plane the directivity of the antenna could be increased and thus the gain. With the finite ground plane the area where the gain is high is then reduced compared to the infinite ground plane case. You have then a more directive antenna at the main beam, but the beam is smaller.

 

[volker_muehlhaus]

I ran the analysis in Sonnet, to compare gain with finite and infinite ground plane. In my model, I used lossy metal instead of PEC.

with low loss substrate tand= 0.002:
Gain with infinite ground plane 5.5dBi
Gain with finite ground plane 5.2dBi

with FR4 loss tand= 0.027:
Gain with infinite ground plane 4.5dBi
Gain with finite ground plane 4.3dBi

 

[Abu Maria.]

thank you volker , thank you all !
so i have right ! gain increase with finite ground plane !
I thought if we put a finite ground plan , it'll give back lobe therefore reduction of the gain,but i see that give inverse , gain is increased!!!

physically how we explain this!?

thank you very much!

 

[volker_muehlhaus]

I thought if we put a finite ground plan , it'll give back lobe therefore reduction of the gain,but i see that give inverse , gain is increased!!!

In my results, the gain with finite ground plane is lower, as expected from your back lobe concept.

 

[ferdows]

thank you volker , thank you all !
so i have right ! gain increase with finite ground plane !
I thought if we put a finite ground plan , it'll give back lobe therefore reduction of the gain,but i see that give inverse , gain is increased!!!

physically how we explain this!?

thank you very much!

i dont have believe to the result of ADS for antenna simulation as i told but if u think ADS work truly so your problem many be oriented from some your change in simulation condition

 

[Abu Maria.]

ah!thank you volker ! also thank you ferdows!
please tell me
dimension of ground plane , in general , how choose the size of ground plane ? and how this finite ground influence in parameter s11 so in resonate frequency!

 

[volker_muehlhaus]

dimension of ground plane , in general , how choose the size of ground plane ? and how this finite ground influence in parameter s11 so in resonate frequency!

The size of your ground plane is fine, no problem with that. The rule with margin = 3...4x substrate height makes sense.

I think that you have a problem from using different port types in Momentum, which leads to these differences in results. I have more trust in the Momentum simulation with infinite ground, because that is the "default" case which has been used and tested for many years, and the port calibration method is straightforward.

For the finite ground, there might be some error introduced by the large distance between the signal and ground reference port. If you look at the Momentum document, you will see warnings about accuracy of these ports if there is too much distance between signal and ground reference. For my simulations in Sonnet, I used the same port type in both cases (via port) and the S11 was almost identical for finite and infinite ground. From that, I think that if you see large differences in your Momentum model, the differences are from different port types and not from the size of the ground plane.

I don't know if you really need to simulate with finite ground, to see the back lobe. Maybe it is better to do the "normal" model, with infinite ground, as most people have used it for many years. That should be "safe" for the port configuration, and is also much more efficient for meshing and simulation time. The change in gain from the finite ground size is small, as you can see from my Sonnet results above.

 

[Abu Maria.]

thank you!