Lumped Port vs. Impedance Boundary

[margarida]

When I simulate my patch antenna using impedance boundary vs. lumped port with the same impedance, I get different results.

Which result should i trust?

 

[PlanarMetamaterials]

Where is this boundary/port located in your simulation domain? Is this using HFSS? Why are there scattering data present for three ports?

 

[margarida]

Yes, in HFSS. I have a patch that has 4 ports. I will feed two of the ports, 90º phase difference between them, to emulate circular polarization. I scattered data for 3 ports because: I want to see if the antenna is tuned through S11 and see cross-talk between ports using S(3,1) and S(4,1).

I have attached an image for better explanation of the geometry. I have suppressed the meander lines, to cut a few variables of my tuning, that's why I insert ports directly on the patch antenna.

 

[PlanarMetamaterials]

Hmm, ok. One discrepancy I can think of is the number of modes allowed on the surface. I believe the lumped port will only allow the fundamental mode to exist, whereas the impedance boundary might allow for any mode. This is just a guess, but if correct, it would mean that the impedance boundary is more trustworthy.

However, since the lumped surfaces are very close to the patch, likely they will be interacting with some of the fields from inside the patch. I think the optimal simulation methodology here would be to increase the length of the feedlines, and observe how the responses change.

I don't have access to HFSS to check, but if you can increase the number of modes on the ports, try increasing the count and observe how this influences the results.

 

[margarida]

Thanks for your reply. It seems I can't add more than one mode on a lumped port, couldn't figure out why. I know that on wave ports I can add up to 25 modes, however wave port is an excitation used when physical size of the port is comparable to or larger than a wavelength, which is not the case here.

However, I increased the length of the feedlines (3mm, roughly a wavelength), and observed that the resonance frequency shifted to the right. However, when Impedance Boundary was used, the shift was to the left. Am I missing something here?

 

[volker@muehlhaus]

It seems I can't add more than one mode on a lumped port, couldn't figure out why.

A lumped port has plus and minus terminal, and is defined by voltage and current between the two terminals. The terminals themselves are considered small compared to wavelength, so the lumped port only supports one mode.

I don't understand your modelling approach, but wonder if the distributed boundary touching the antenna along the entire side vs. lumped (!) port at one specific location is the reason for the difference.

 

[margarida]

I don't understand your modelling approach,

Why?

distributed boundary touching the antenna along the entire side vs. lumped (!) port

I use the same rectangle to assign both lumped port or impedance boundary

 

[volker@muehlhaus]

Why?

I have simulated many antennas, but never saw patch antenna simulation before with ports on all sides, so I don't understand the purpose of this approach.
For dual polarisation, I would expect two ports only, that is sufficient to analyze cross talk.

 

[margarida]

The idea is to design a series array, that's the reason of the other ports. I need to tune how much power is being transferred to the next patch, phase delay, etc.

 

[PlanarMetamaterials]

Moving the boundaries further away shouldn't really cause any frequency shifting at larger distances. I might suggest starting by validating the numerical setup. Run the simulation with different amounts of conversion/initial mesh sizes, and confirm that you get the same results in all cases.