FEKO : wave through a really thin plate (thickness around skin depth)

Hello,

I am trying to use FEKO to solve a problem where I basically have a plate whose thickness is less than the skin depth of my metal at the frequency of interest :

freq : 1GHz
skin depth : 2um
material : copper
thickness of plate : 500nm

I used the 3d MoM method since the 2d method cannot converge here (I have an other post about it but basically in 2d, feko asks for a thickness of metal and it has to be greater than skin depth AND smaller than triangle mesh size, which is 1) absurd when thickness is smaller than skin depth and 2) impossible when triangle mesh becomes smaller than skin depth)

So I have two questions :

1. Is the 3d MoM correctly representing the wave "passing through" my plate ? I model the copper by specifying "no thickness" to my faces

2. Should I use FEM instead to model accurately this ? I tried but FEKO tells me FEM applies only to dielectric, which is not my case since I got a metal !

Thanks a lot,

Pierre

Find more posts tagged with

Feko

Accepted answers

All comments

degleve_jarmlami

2. In addition, when trying to use FEM on a dielectric whose conductivity represents copper, the number of tetrahedron is absurdly high ! like for 1 Tb of mesh. Dimension of my plate are however resonable : 20mm * 4 mm * 0.5mm, and wavelength is around 300mm anyway so I was expecting reasonable mesh of around 1mm.

Is it because I try to represent metal as dielectric ? I've noticed using an other dielectric with no conductivity it asks for less tetrahedron …

Mahan Rudd_22044

Hi @degleve_jarmlami,

When you say "Feko asks for a thickness of metal and it has to be greater than skin depth" where did you get this from? I mean did you get any warning stating the same thing or you saw that in the manual or somewhere else. I ran a simulation with the parameters you mention on the same model size (cuboid of 20mm x 4mm x 0.5mm), free space region, faces assigned copper with variable thickness (through parameter sweep) of 50nm, 500nm, 2um and 10um and looked at the loss calculation and it seems to be calculated properly varying from higher losses for thickness smaller than the skin depth and reaching to a constant value as the thickness goes beyond multiple skin depth.

Also note that the model size is too small compared to the wavelength at 1 GHz (30cm). You need to set a custom mesh size (something like 0.5mm should be good).

By the way, what is the objective of your simulation? are you looking at metallic loss calculation or shielding effectiveness?

Feel free to share your model if possible and I will have a closer look.

Regards

Mahan

degleve_jarmlami

Hello, thanks for getting in touch.

. "where did you get this from" : I find it in the documentation of FEKO. (see SK card sections in the User Guide)

. Concerning your simulation : I made a mistake in my measurement. I meant a thicknes of 0.5um and not 0.5mm. I want to have a thickness inferior than my skin depth.

—> For the loss calculation, I see no problems. Problem is if I have a thickness smaller than the skin depth, will the wave effectively go through ?

Since feko asks for a thickness greater than the skin depth, I guess it is because it does not compute the passing wave, right ?

Thanks again,

Pierre

Jaehoon

Hello @degleve_jarmlami,

Regarding your answer, ""where did you get this from" : I find it in the documentation of FEKO. (see SK card sections in the User Guide)",

→ Your answer was probably based on the following sentence from the page 1389 of the Feko user guide. (The surface impedance represents a metallic face. I mean this is a boundary condition.)

However, when the skin depth is larger than the metallic face's thickness, Feko uses the following surface impedance automatically, which is given at the page 1390.

Therefore, the Feko's MoM and FEM solver has no issue with various metallic face's thickness which is smaller or larger than the skin depth, as far as I know.

I mean you can compute the propagating wave through a metallic face whose thickness is smaller than the skin depth with Feko.

Best regards,

Jaehoon

degleve_jarmlami

Hello,

Thanks for you're really argumented answer. Indeed I just saw exactly this yesterday and thought I might have solved the problem as well ! I am gonna make some test and let you know :)

Last question I have : there is an other condition of FEKO that imposes the specified thickness to be smaller than the mesh triangles sizes. This makes some troubles for me since I try to resolve thin details (triangles <0.1um) with thickness not that small (~0.5um). This raises "WARNING 2757: A triangle is too thick as compared to the lateral dimensions for application of the skin effect approximation"

How to handle this situation in your mind ?

Thanks for your support,

Pierre

Jaehoon

Hello Pierre,

To resolve the warning, we normally limit/decrease the simulation upper frequency.

If you share your model (.cfx file) here, let me investigate how to resolve it in another way.

(If your model has a security issue, you might modify/simplify it before sharing.)

Best regards,

Jaehoon

degleve_jarmlami

Hello, thanks for answering.

Why limiting the upper frequency ? I can share a model if needed, but really what happens is simply that I want to have triangles whose size is smaller than the thickness !

Pierre

Jaehoon

Why limiting the upper frequency ?

→ If you decrease the upper frequency, the edge length of the triangles becomes larger, so you can resolve the warning. I believe this is not what you want. So I asked you to share your model.

Regards,

Jaehoon

degleve_jarmlami

Okay, I see.

Here is a quick model that got the problem. This is a square plate, with a slot of 1um gap. Here, triangles needs to be of 1um/10 size to capture phenomena. However, thickness is 1um.

—> This gives a thickness too big for the triangles !

degleve_jarmlami

i share a rar since cfx does not work for some reasons

example.rar

Jaehoon

Hello Pierre,

After checking your model, I realized you didn't have to use cuboids to draw your geometry.

You can simply construct two rectangular faces (with thickness) instead of two cuboids, since you are using MoM (Method of Moments).

2. However, because the gap of the subtracting face is the same as the thickness, Feko generates the same warning massages from the triangle (ID=37) one of whose edges is the gap. I think it cannot be escapable due to the gap dimension.

However. s

ince the face's thickness is larger than the skin depth and the warning message is only from one triangle, the solved results look acceptable.

Please see the attached model and let me know if you have further question.

example_mod.zip

Best regards,

Jaehoon

degleve_jarmlami

Hello,

Yes indeed, that's what I think about it, if even 10% of my triangles become PEC, it would be still good probably …

But you confirm there is no workaround this problem ? Even in 3D MoM I still have this Warning issue !

Jaehoon

Hello Pierre,

Since the skin depth is ~3.7 times smaller than the specified surface thickness (according to the out file as below), you can safely ignore the warning. Feko will still apply the same conductor impedance onto the small triangle(37).

2. The warning is mainly to ask the user to double check if the thickness is set not too large compared to the mesh size mistakenly.

Best regards,

Jaehoon

degleve_jarmlami

Thanks for answering.

I understand ! So you mean the triangles are not turned into PEC triangles. I thought it was the meaning of "All segments and metallic triangles without a listed label are perfectly conducting."

My bad.

So can you confirm that :

If I happen to have a thickness which becomes smaller than my skin depth (which is going to happen in my project later), I must simply change the skin depth card approximation from "High frequency approximation" to "Static approximation" ? I can read in the doc they propose opposites conditions on the thickness relative to 2*skin_depth

Thanks a lot, it enlights my simulations !

Pierre

degleve_jarmlami

Hello again,

I have a question regarding your suggested way of building the block with a rectangle and adding a thickness to it.

I've tried a simulation by instead of having a rectangle and sweep it by my thickness value, I keep it flat and just define my thickness if the face property when defining it as metal.

I have really different result and would like to know why is it so ?

Thanks

Jaehoon

Could you share your two Feko models for me to check?

Regards,

Jaehoon

degleve_jarmlami

I got some privacy issues regarding my models.

I'll try make one with the same problem !

Jaehoon

I already compared it in your different question regarding edge meshing…

Best regards,

Jaehoon

degleve_jarmlami

Hello again,

I contact you again concerning the discussed model.

In an other thread, we found out that that near field would be different between a 3d model and a 2d with a thickness applied. However, far fields should look the same.

I can indeed observe this result with some simulations (like the electric dipole antenna discussed in the other thread).

However, in cases like this one, even the far field are different between the two models. In my opinion, it is because FEKO does not compute the kind of capacitance effect that occurs here since the gap is of the same size of the thickness.

What bothers me is that even when I go into far field (100 times the thickness) and make the gap larger than the thickness (50 times), I still got quite different result in terms of resonant frequency of my structure and intensity of field.

Would you know how to proceed in this case ? I add that I specify the same thickness in the metallic faces in both cases. Even dividing by two in the 3d case does not help.

Thanks really much,

Pierre

Jaehoon

Could you please share the Feko models helping me understand your question?

Thank you,

Jaehoon