Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

EM simulator for multiple dielectric layer losses

Status
Not open for further replies.

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
dielectric layer losses

This question goes to all EM wizards.

Lets say we want to simulate and find out what should be the transmission/reflection loss of one or multiple dielectric layers. Which EM simulator do you think will easily do the job and how would you set the following generic (with most simulators) conditions:

1. Boundary conditions should be set to periodic, open, or electric? We don't care about any "edge" effects of the material but only for the transmission-reflection characteristics of the layer (In a sense the material can be considered to be infinite).

2. Can we excite the layer and find the transmission-reflection losses through waveguide ports and a plane wave excitation? How can we investigate/find in both cases (if applicable) the following:

(a) S21, S11 parameters (or an equivalent way to predict the transmission-reflection losses)
(b) Introduce the effect of the polarisation state of the excitation (Is this possible only with a plane wave excitation?)
(c) Introduce the effect of the angle of incidence of the incident wave (Is this possible only with a plane wave excitation?)


Any input is welcomed on the easiest way to simulate the above, (simulator and simulation setup wise).

Many Thanks
Cross
 

loucy

Advanced Member level 1
Joined
Aug 26, 2001
Messages
414
Helped
24
Reputation
48
Reaction score
11
Trophy points
1,298
Activity points
4,657
you have not stated what you want clearly. The reflection/transmission loss would depend on the excitation. So are you talking about normally incident plane wave, or microstrip line on a multilayer media, or a periodic structure on a layered media....
 

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
Hi loucy,

I am trying to predict the loss suffered(in dB) when a plane wave travels through a single or multi layer dielectric structure (an example can be a plane wave travelling through a single pane or double glazed window. What would be the loss introduced by the window?) The plane wave can be incident at any angle of incidence (as one would expect the loss should change depending on the angle of incidence and the polarisation state).

Best Regards
Cross
 

loucy

Advanced Member level 1
Joined
Aug 26, 2001
Messages
414
Helped
24
Reputation
48
Reaction score
11
Trophy points
1,298
Activity points
4,657
If you don't care about the edge effect, the best way is to write your own program to calculate the transmission and relfected waves. Isn't this a textbook problem?
The usual FEM or FDTD tool (3D) would introduce aritificial boundary acting like some aritificial edge. The 2.5D MoM tools might not be useful, because you don't have a metal, and there is no unknown currents for them to solve for.
Maybe a circuit simulator is useful, you just draw a series of transmission lines (whose characteristic impedances are dependent on the material parameters, incident angle and polarization). You can find the relation on many textbooks. In this case, you can get some S-parameter output from commerical tools directly.
 

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
Thanks loucy,
The problem is not the simplified single or multilayer dielectric structure. I used this description as an example to give people an idea of what I am after. What I really want to do is to simulate bricks with air holes and concrete with metal grids (i.e periodic structures). This is where I thaught a 3D em simulator would come handy instead of trying to code the problem myself.

Any thaughts?

Cross
 

loucy

Advanced Member level 1
Joined
Aug 26, 2001
Messages
414
Helped
24
Reputation
48
Reaction score
11
Trophy points
1,298
Activity points
4,657
The best choice might be HFSS.

Some FDTD softwares also have periodic boundary conditions, but I am not sure about the accuracy of the algorithm. Can anyone comment on this? One is called the split field formulation?

IE3D 9 also has option for setting periodic boundary, but it is not clear to me how they have implemented it. They give an option for setting the number of images, so it appears that the green's function used is not for periodic structure. It takes a lot of time to run if you set the number of image too high (like 20). So the accuracy might not be good enough.

Another MoM code, EmPicasso (?) can also simulate periodic structure, but the version I tested (2 years ago) can't have metal touching the periodic boundary. So it is not useful for your case.

Although I think HFSS is the best choice, I have never seen an accuracy test on it. You might also check out the FEMLAB, another code that uses finite element method.
 

mwpro

Member level 4
Joined
Apr 20, 2001
Messages
71
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,288
Activity points
700
This subject raise a good issue. Which simulator can simulate plane wave? HFSS cannot.
 

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
CST can excite a structure with a plane wave but I think you cannot use it when the boundary conditions are set to periodic.
 

loucy

Advanced Member level 1
Joined
Aug 26, 2001
Messages
414
Helped
24
Reputation
48
Reaction score
11
Trophy points
1,298
Activity points
4,657
Give your argument for stating HFSS can't handle plane wave. Simply draw a cube, then you can select a face and assign plane wave source. If you are talking about edge effect, accuracy, or whatever, then it is a different issue.
 

totokevin

Newbie level 6
Joined
May 27, 2003
Messages
13
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,283
Activity points
126
Could you describe your structures more clearly? Is it a PBG or a FSS? From my point of view, FEM is suitable to PBG, while MoM is a good choice for FSS. Ansoft HFSS and Designer(Ensemble) are ideal to get Tx/Rx. FEMLAB is powerful in eigenmode analysis. BTW, although I have never used CST MWS, I know that it supports the periodic boundary conditions.
 

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
The 2 structures that interest me are the following:

1. A brick wall where bricks are attached together with a thin layer of mortar.
2. A concrete wall incorporating a metalic mesh (grid).
 

goxy

Full Member level 4
Joined
Jul 25, 2002
Messages
234
Helped
6
Reputation
12
Reaction score
5
Trophy points
1,298
Activity points
2,283
In CST MWS can be set periodic boundary phase shift (Solve>Boundary Condition>Periodic Boundary Phase Shift),
but first in Boundaries for axis of interest must be set Periodics.
And can be applies only to the frequency domain solver and the eigenmode solver, not to the transient.
See helix example.

**************
Eigenmode Analysis>SlowWave>helix
 

totokevin

Newbie level 6
Joined
May 27, 2003
Messages
13
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,283
Activity points
126
The 1st structure is a PBG crystal. The 2nd one is a frequency selective surface with dielectric layers. These periodic structures have been analyzed using MOM (Mittra, PSU), FDTD (Samii, UCLA), FEM(Volakis, UMich), or TLM (Parker, Kent). The related papers can be easily found in IEEE Trans. AP/MTT.
 

totokevin

Newbie level 6
Joined
May 27, 2003
Messages
13
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,283
Activity points
126
Simulating software for these approaches are:
Ansoft HFSS (FEM), Ansoft Designer/Ensemble (MOM), XFDTD (FDTD), CST MWS (FIM), Micro-stripes (TLM).
 

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
Thank you totokevin. Do you have any experience with these simulators concerning the above strucrtures?

Cross
 

totokevin

Newbie level 6
Joined
May 27, 2003
Messages
13
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,283
Activity points
126
Generally, the frequency domain methods (MOM/FEM) lead accurate results, but they are time consuming. The time domain methods (FDTD/FIM/TLM) cost much less calculating time, while they can not guarantee the accuracy when analyzing some irregular or complex geometries. The choice of the approaches usually depends on your requirement.
 

totokevin

Newbie level 6
Joined
May 27, 2003
Messages
13
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,283
Activity points
126
No matter which method is used, the key to satisfying results is drawing your structure correctly and setting the right boundary conditions. How ever, according to my experience, if I were you I prefer FEM for structure 1 and FDTD/FIM for structure 2.
 

cross

Member level 4
Joined
Dec 31, 1999
Messages
72
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
509
Thank you totokevin. let's say I have drawn the structure correctly. How would you go forward to measure the transmission and reflection loss. With waveguide ports or plane wave excitation? How would you also setup the boundary conditions?

Many Thanks
Cross
 

loucy

Advanced Member level 1
Joined
Aug 26, 2001
Messages
414
Helped
24
Reputation
48
Reaction score
11
Trophy points
1,298
Activity points
4,657
How is a PBG structure different from a Frequency selective surface? What is the basis for choosing FEM for PBG, and FDTD for FSS?
 

totokevin

Newbie level 6
Joined
May 27, 2003
Messages
13
Helped
2
Reputation
4
Reaction score
2
Trophy points
1,283
Activity points
126
Hi Cross,

The Tx/Rx can be derived by solving the integral/differential equations. The excitation and boundaries depend on your structures.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top