Menu

Discussion Forum

New Discussion
Filtrer

Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

PML and RF Module

Posted 29 sept. 2010, 14:17 UTC−4 RF & Microwave Engineering, Wave Optics Version 4.0a 12 Replies

Tobias Zederbauer
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Hi,

I am trying to find the eigenmodes of a grated ring waveguide. To do so I simulate a segment of the ring and use periodic boundary conditions. Since light should be radiated upwards and downwards PMLs should be added to avoid reflections. However after adding the PMLs the solver always crashes, giving the message:

-------------------------------------------------------------------------------------------------------------------------
Exception:
com.femlab.jni.FlNativeException: Undefined value found Solvermodel
Messages:
Undefined value found Solvermodel

Stack trace:
at solvermodel.cpp, row 1837, (Undefined value found in the stiffness matrix.
There are 6328 equations giving NaN/Inf in the matrix rows for the variable tExEyEz10
at coordinates: (0.000189496,-1.0796e-06,1.474e-05), (0.000189496,-9.39261e-07,1.474e-05), (0.000189696,-1.07929e-06,1.474e-05), (0.000189696,-9.37368e-07,1.474e-05), (0.000189563,-9.3969e-07,1.474e-05), ...
There are 4445 equations giving NaN/Inf in the matrix rows for the variable tExEyEz20
at coordinates: (0.000189496,-1.00943e-06,1.4809e-05), (0.000189563,-1.08003e-06,1.474e-05), (0.000189629,-9.38529e-07,1.474e-05), (0.000189746,-1.00978e-06,1.474e-05), (0.000189846,-8.70737e-07,1.474e-05), ...
There are 4445 equations giving NaN/Inf in the matrix rows for the variable tExEyEz21
at coordinates: (0.000189496,-1.00943e-06,1.4809e-05), (0.000189563,-1.08003e-06,1.474e-05), (0.000189629,-9.38529e-07,1.474e-05), (0.000189746,-1.00978e-06,1.474e-05), (0.000189846,-8.70737e-07,1.474e-05), ...
and similarly for the degrees of freedom, NaN/Inf in the matrix columns
)
at com.femlab.solver.FlSolver.femEig(Native Method)
at com.femlab.solver.FemEig.run(Unknown Source)
at com.femlab.server.FlRunner.run(Unknown Source)
at com.femlab.util.i.run(Unknown Source)
at com.femlab.util.aa.run(Unknown Source)

at com.femlab.solver.FlSolver.femEig(Native Method)
at com.femlab.solver.FemEig.run(Unknown Source)
at com.femlab.server.FlRunner.run(Unknown Source)
at com.femlab.util.i.run(Unknown Source)
at com.femlab.util.aa.run(Unknown Source)

-------------------------------------------------------------------------------------------------------------------------------------------

I dont know what to do about that, maybe someone could give me a hint

thanks in advance
tobias
12 Replies Last Post 18 mai 2013, 00:19 UTC−4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 sept. 2010, 14:51 UTC−4
Hi

have you checked the material data sub-node ? when you add a PML yu must also redefine the material data use, by default it's in epsilon_r mu_r and not n,k, even if the main physicshas been redefined to n,k.

This is just one possibility, might not be the issue ;)

--
Good luck
Ivar
Hi have you checked the material data sub-node ? when you add a PML yu must also redefine the material data use, by default it's in epsilon_r mu_r and not n,k, even if the main physicshas been redefined to n,k. This is just one possibility, might not be the issue ;) -- Good luck Ivar
Tobias Zederbauer
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 sept. 2010, 15:32 UTC−4
Hi,

I'm not sure if i got what you meant, in the subdomain window, at the physics tab i use
n to define the material.

best regards
Tobias Zederbauer
Hi, I'm not sure if i got what you meant, in the subdomain window, at the physics tab i use n to define the material. best regards Tobias Zederbauer
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 sept. 2010, 17:00 UTC−4
hi

yes I expected that, but there is one in the PML definition sub-node too and both/all must probably be set to n,k and not left to their default eps_r mu_r. I have been fooled a few times already on that one ;)

--
Good luck
Ivar
hi yes I expected that, but there is one in the PML definition sub-node too and both/all must probably be set to n,k and not left to their default eps_r mu_r. I have been fooled a few times already on that one ;) -- Good luck Ivar
Tobias Zederbauer
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 sept. 2010, 10:39 UTC−4
Hi,

I'm still not sure what you mean, all subdomains that should be PMLs have the same settings as the bordering subdomains. i.e. n=1 on top since the PML is connected to the air box, and n=3.05887940239403+i*0.0009945241056400001 at the bottom since this PML is connected to the substrate.

however the problem does not appear when I solve for harmonic propagation, only when searching for eigenvalues.

tobias
Hi, I'm still not sure what you mean, all subdomains that should be PMLs have the same settings as the bordering subdomains. i.e. n=1 on top since the PML is connected to the air box, and n=3.05887940239403+i*0.0009945241056400001 at the bottom since this PML is connected to the substrate. however the problem does not appear when I solve for harmonic propagation, only when searching for eigenvalues. tobias
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 oct. 2010, 03:25 UTC−4
Hi

As I said above, but I did not expalin it well enough, n,k must be set in TWO different places for RF, in the main "Wave equation" setting (capture.png) AND in the Wave Equation sub-node of the PML node (capture1.png)

This is easily forgotten, but if you have set both correctly, then you have another isue, that I do not understand ;)

--
Good luck
Ivar
Hi As I said above, but I did not expalin it well enough, n,k must be set in TWO different places for RF, in the main "Wave equation" setting (capture.png) AND in the Wave Equation sub-node of the PML node (capture1.png) This is easily forgotten, but if you have set both correctly, then you have another isue, that I do not understand ;) -- Good luck Ivar

Attachments:

Tobias Zederbauer
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 oct. 2010, 06:22 UTC−4
Hi,

I use COMSOL 3.5a so my GUI is quite different, and I don't know COMSOL 4.0 but from what I see on your pictures, I think my settings are the same.

tobias
Hi, I use COMSOL 3.5a so my GUI is quite different, and I don't know COMSOL 4.0 but from what I see on your pictures, I think my settings are the same. tobias
Yassin Chowdhury
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 19 mars 2011, 19:22 UTC−4
Hi,

1. In the dielectric_scattering_PML.mph model file given as example by COMSOL, they use 3 different wave equations for different layers of PML i.e. one equation each for horizontal, vertical, and corner layers of PML(cartesian/rectangle).
Is it really necessary or can I use just one wave equation for the whole PML ? (i'm using 4.0a)

2. What should be the optimum thickness of the PML layers for the case of plasmonic based waveguide. Do they still need to be 'lamda0=1.55um' thick ?

3. When I have multilayered medium(Metal/Si/Substrate), should I use just the outermost material as PML, or do I need to define air surrounding, then apply PML ?

4. How to find the optimum size of total domain area, isn't it true if I apply PML, I can use smaller domain to reach the correct answer ?

5. Should I use scattering boundary condition even after using PML ?


Thanks in Advance,

-Yassin.
Hi, 1. In the dielectric_scattering_PML.mph model file given as example by COMSOL, they use 3 different wave equations for different layers of PML i.e. one equation each for horizontal, vertical, and corner layers of PML(cartesian/rectangle). Is it really necessary or can I use just one wave equation for the whole PML ? (i'm using 4.0a) 2. What should be the optimum thickness of the PML layers for the case of plasmonic based waveguide. Do they still need to be 'lamda0=1.55um' thick ? 3. When I have multilayered medium(Metal/Si/Substrate), should I use just the outermost material as PML, or do I need to define air surrounding, then apply PML ? 4. How to find the optimum size of total domain area, isn't it true if I apply PML, I can use smaller domain to reach the correct answer ? 5. Should I use scattering boundary condition even after using PML ? Thanks in Advance, -Yassin.
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 mars 2011, 05:02 UTC−4
Hi

a few comments, describing "my way":

for PML you need to define corectly the direction of propagation, otherwise the physics are wrong and the extension direction no good, hence no effect or convergion issues. Personally I prefer to use spherical/circle items and IMPORTANT define correctly the centre !

for the thickness I use typically 1/5 or 1/10 of the total domain size.

for the material properties, I use the same as adjacent boundary in the propagation direction. So sometime I cut my PML to respect the adjacent material combinations

--
Good luck
Ivar
Hi a few comments, describing "my way": for PML you need to define corectly the direction of propagation, otherwise the physics are wrong and the extension direction no good, hence no effect or convergion issues. Personally I prefer to use spherical/circle items and IMPORTANT define correctly the centre ! for the thickness I use typically 1/5 or 1/10 of the total domain size. for the material properties, I use the same as adjacent boundary in the propagation direction. So sometime I cut my PML to respect the adjacent material combinations -- Good luck Ivar
tianrun chen
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 18 août 2011, 13:00 UTC−4
Hi Tobias,

I am wondering if you have solved the PML problem by now. I have exactly the same problem with you.

thanks,

Tianrun



Hi,

I use COMSOL 3.5a so my GUI is quite different, and I don't know COMSOL 4.0 but from what I see on your pictures, I think my settings are the same.

tobias


Hi Tobias, I am wondering if you have solved the PML problem by now. I have exactly the same problem with you. thanks, Tianrun [QUOTE] Hi, I use COMSOL 3.5a so my GUI is quite different, and I don't know COMSOL 4.0 but from what I see on your pictures, I think my settings are the same. tobias [/QUOTE]
Akram Hajebifard
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 18 avr. 2012, 12:03 UTC−4
Hi Yassin,

Could you find answer for any of your questions?
I'll appreciate it if you could share your ideas with us.

Hi Yassin, Could you find answer for any of your questions? I'll appreciate it if you could share your ideas with us.
Petr Donzov
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 4 mai 2012, 08:05 UTC−4
May be I have similar problem. May be not. I have just started working in Comsol and now I am trying to do simple eigenvalue simulation with RF Module. Just square with n2 surrounded by pml with n1.
If I disable PML node than obviously everything is ok.

The error is: Undefied value found. What can be the problem?
May be I have similar problem. May be not. I have just started working in Comsol and now I am trying to do simple eigenvalue simulation with RF Module. Just square with n2 surrounded by pml with n1. If I disable PML node than obviously everything is ok. The error is: Undefied value found. What can be the problem?
Lujun Huang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 18 mai 2013, 00:19 UTC−4
Hi, I just have the same problem as you. If you have solved this problem, could you by any chance tell me how to solve this problem. Many thanks

Lujun
Hi, I just have the same problem as you. If you have solved this problem, could you by any chance tell me how to solve this problem. Many thanks Lujun

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.

Suggested Content