UC-EBG for patch antennas

Hello, I want to insert some uni-planar EBG structure between two patch antennas to supress the surface wave, and thus reduce coupling effect.

My problem is, that I'm not abble to see the surface wave excitation effect. Even if i delete the substrate between antennas, S21 and S11 are not being changed.

Is there a way to enhance the surface wave effect? Or am I just probably doing something wrong with the simulation? What should be the right simulation settings?

I'm using CST MWS for simulation. I tried multiple frequencies between 2 GHz and 5 Ghz.

Hi raf_prze,

I've done a great deal of research on this, but I use HFSS instead of CST. Can you explain your simulation setup? I still might be able to help you.

Thanks! I really have some hard time with that stuff.

My simulation setup is quite basic. I've got two antennas, and I compare results from model with and without EBG. I just focus on S21, using time domain solver. I mostly use standard mesh density values etc. Should I make this more complex? I'm quite new to this kind of software. I would be very grateful for some useful tips.

Maybe the surface wave is not so significant in my case. I know that it depends on frequency, substrate height and dielectric constant, but I would like to keep it low-cost. Preferably poor FR4...

I also have simulated some EBG structures individually, and they work well (-50dB in stop band).

Hi raf_prze,

I'm not really sure what setup tips I can give due to the fact that I've never used CST. Just make sure that your mesh in the air gaps between unit cells is at least 2 tetrahedrons across each gap.

You should be able to see the reduction due to surface wave supression between two antennas. I can confirm that I've done these simulations and they work well. What frequency band are you working in?

An important issue is how you simulate the EBG structures. IMO, many of the simulation methods reported in the literature will categorically give incorrect results. The fact that you're seeing -50 dB in the stop band suggests to me that your setup is invalid - although this depends on what frequency you're working at.

Well... 2.4 GHz would be best for me, but I'm considering some higher, due to dimensions. Yeah, I know that these are not practical frequencies, but this is supposed to be didactic-like example.

When I got -50dB, I put ports directly on electric surface of EBG on both sides. I am aware of fact, that this won't be near that in full model with gaps, but I just checked it only for right frequency band , which, what I know, is relevant to cell structure (capacitance and inductance) and is quite constant.

What was your S21 reduction in your simulations? Have you done some projects with some "low" frequencies?

I don't really understand your setup - I'm assuming you're looking to suppress the TM0 mode surface wave due to the frequency (the others are below cutoff unless you have a really fat dielectric). Are you inducing currents on the EBG surfaces? And are the UC-EBGs backed by a ground plane on the other side of the substrate? What is above the UC-EBGs?

I'm actually using them at "low" frequencies: L-band. At these frequencies, I'm lucky to see 1dB of suppression per unit-cell. At 2.4 GHz on FR-4 you're probably going to get a maximum of 2.5dB per unit-cell.

Of course, there is solid ground plane. There is nothing above EBG. It is on the same substrate plane as patch antennas. Can you maybe propose some cell designs?

2.5 dB would be great... My best "score" so far is 0.4 dB.

I have some simulation files but they are all HFSS, and they are not for that frequency range, sorry.

I am curious as to how you got a result of -50dB - based on what I've previously seen, I would guess that your transmission simulation setup is shifting the band gap of the structure from what it is in the antenna setup, hence your not so great results. What do you mean by there is "nothing" above the EBG? Do you have any boundary condition being enforced at any point above the surface? If you have a picture of your setup, that would be useful.

Boundary conditions are open (add space), except ground plane, which is electric

https://i.imgur.com/MzAcYvZ.png
https://i.imgur.com/mU1fXtk.png

In this model i used discrete ports. Also, the EBG structure on picture is for example only.

Yes, that setup looks correct - however, I was more curious about the simulation in which you received -50 dB (when you "simulated some EBG structures individually"). The reason is that this setup, rather than your antenna setup, may be giving you incorrect results.

Well, I just wanted to check frequencies of the stop band defined by cell. The drop in decibels didn't mean much in that simulation.

https://i.imgur.com/ObFdei6.png
or
https://i.imgur.com/O2z3EqL.png


Still, I think my main problem is the lack of surface wave. Deleting part of substrate (with ground plane) between antennas doesn't seem to change anything.

Unfortunately, I don't know enough about CST to comment on the setups, but as long as those are absorbing boundaries you should be good. The surface waves will be created by the antennas - I don't think you need to worry about that. See Yang and Rahmat-Samii if you haven't already. Remember that at such low frequencies, the surface wave is so loosely bound that the dielectric hardly makes a difference.

Edit: (Hardly makes a difference in terms of the surface wave properties. Obviously it will affect the EBGs and Antennas to a much larger degree.)

- - - Updated - - -

I think my only suggestion at this point would be to ensure that everything is the same between the two simulations. A different level of meshing for example, could cause the resonance frequency to shift.