AnsweredAssumed Answered

Momentum vs FEM: what's the problem ?

Question asked by Fresnel on Aug 17, 2018
Latest reply on Sep 19, 2018 by Riju

Like • Show 0 Likes0
Comment • 10

Hello,

I would like to apologize for this question since I am quite it's a recurring question...

I design an antenna (J-dipole) using Momentum: results are great...

However, I would like to check the simulation with FEM to take account of finite FR4 substrate...

1) with the default options, results with FEM are odd: the question is how to set the FEM properly? I read from documentation and it is not very clear...

2) one FEM siting will work, how to "cut" the substrate ? With 3D visualisation tool, the FR4 layer is always the same size of simulation_box...

Thanks for your reply.

Attachments

No one else has this question

Outcomes

10 Replies

volker_muehlhaus Aug 17, 2018 10:29 AM
Mark Correct
Correct Answer
Have you set enough spacing to the simulation boundary, and open boundaries?

I think you can use the "boundary area layer" setting to specify a layer that determines the substrate boundary for FEM.
Myself, I use this method: For finite dielectrics, I specify air for the base dielectric at that layer, and insert a dielectric via that represents the PCB material.
Like • Show 1 Like1
Actions
Fresnel Aug 17, 2018 2:59 PM
Mark Correct
Correct Answer
Thank you very much for your answer: boundary area layer makes the difference.
However, how to improve the solution ?
Is the difference between two simulations comes really from the finite substrate ?
Or I miss something in the definition of port or in FEM options ?

I put my files.
About FEM: your can see what I change.
The only other change is the convergence: I put 2 for "Consecutive passes of delta error required"

Any suggestions for improvements ?
(I am little bit surprise by the difference...
Thanks a lot
F.

FEM

MoM
Like • Show 0 Likes0
Actions
- volker_muehlhaus @ Fresnel on Aug 19, 2018 1:04 AM
  Mark Correct
  Correct Answer
  I can't see this from your screenshot: What is the boundary type and boundary distance on the side with the port?
  
  I would prefer to have an absorbing boundary there, and use an internal lumped port for the feed. Distance to the absorbing boundary at least 1/4 wavelength on all sides.
  
  For the "Consecutive passes of delta error required", I usually use 3 to make sure the algorithm has converged and donesn't stop too early at a local minimum.
  Like • Show 0 Likes0
  Actions
  - Fresnel @ volker_muehlhaus on Aug 20, 2018 1:50 PM
    Mark Correct
    Correct Answer
    Hello,
    
    I tried to answer you on Keysight forum, however, it doesn’t work… We have some problems with our network…
    
    So I answer below with new questions.
    Thanks for your time.
    François
    -
    François Boone, ing., Ph. D.
    
    Professeur,
    Directeur du département de génie électrique et de génie informatique
    Université de Sherbrooke
    2500, boul. de l'Université
    
    Tél. : 819 821 8000 poste 65900
    Fax : 819 821 7937
    Courriel : infogegi@usherbrooke.ca<mailto:infogegi@usherbrooke.ca>
    www.usherbrooke.ca/genie<http://www.usherbrooke.ca/genie>
    
    Le 18 août 2018 à 08:07, volker_muehlhaus <moderator_forums@keysight.com<mailto:moderator_forums@keysight.com>> a écrit :
    
    Keysight Community
    Re: Momentum vs FEM: what's the problem ?
    reply from volker_muehlhaus in EDA Software - View the full discussion
    I can't see this from your screenshot: What is the boundary type and boundary distance on the side with the port?
    Boundary type: open (I suppose it’s equivalent to absorbing boundary since the other choices are PEC or PMC.
    
    I would prefer to have an absorbing boundary there, and use an internal lumped port for the feed. Distance to the absorbing boundary at least 1/4 wavelength on all sides.
    I used 25mm on all sides and decide to increase this distance to 50 millimeterI work around 2.4 GHz: let’s say 3 GHz: the free space lambda is then 100mm and the guided lambda < free space lambda.
    
    For the "Consecutive passes of delta error required", I usually use 3 to make sure the algorithm has converged and donesn't stop to early at a local minimum.
    Thanks for this advice.
    
    As you can see in FEM-MoM-Results.png file, I made different simulations with previous parameters.
    You speak about internal lumped port… Since I have the choice between auto or direct port, I assume that these are lumped ports: the two cases give the same result, the blue curve… I don’t trust this result since the design isn't done for resonant antenna (very small bandwidth).
    If I use wave port with TML, zero length, I obtain pink or fuchsia curve: It is a little bit better but not like MoM results.
    
    I check about port: I attach the file 3D-port1.png: I am a little bit surprise by the shape of this port: It seems to me that the shape have to be rectangular, just below micro strip line…
    But it is not the case and I didn’t find any way to change the shape.
    
    Thank you very much if you can give me more advice.
    
    Best regards.
    François
    
    Reply to this message by replying to this email, or go to the message on Keysight Community
    Start a new discussion in EDA Software by email or at Keysight Community
    Following Re: Momentum vs FEM: what's the problem ? in these streams: Inbox
    This email was sent by Keysight Community because you are a registered user.
    You may unsubscribe instantly from Keysight Community, or adjust email frequency in your email preferences
    Attachments
    3D.png179.7 KB
    3D-port1.png212.9 KB
    FEM-MoM-Results.png360.3 KB
    Like • Show 0 Likes0
    Actions
    - volker_muehlhaus @ Fresnel on Aug 20, 2018 11:16 PM
      Mark Correct
      Correct Answer
      Hello François,
      
      you have one edge where both signal and ground line end, and it seems that ADS uses the entire edge length of the ground plane for the port reference width (instead of just a part of the edge underneath the signal line). In Momentum that causes trouble, in FEM I don't know. Can you create an archived ADS project with this antenna and upload it here?
      
      Best regards
      Volker
      Like • Show 0 Likes0
      Actions
      - Fresnel @ volker_muehlhaus on Aug 21, 2018 9:29 AM
        Mark Correct
        Correct Answer
        Ce message vous est transmis par François Boone à l'aide du service d'envoi de
        fichiers de l'Université de Sherbrooke.
        
        Numéro de suivi: 624fab0464
        
        Sujet : Re: - Re: Momentum vs FEM: what's the problem ?
        
        Message : You can download the archive of the project. Best regards.
        
        Télécharger les fichiers
        Nombre de fichiers disponibles: 1
        Expiration: 2018-08-24
        
        Lien: https://www.usherbrooke.ca/envoi-de-fichiers/index.php?token=624fab0464&key=12017deb254ed40ac51b23ce6c4ece6471ee8702
        Like • Show 0 Likes0
        Actions
      - Fresnel @ volker_muehlhaus on Aug 21, 2018 9:30 AM
        Mark Correct
        Correct Answer
        Hello Volker,
        
        My archive is more than 170Mo.
        So I use the unit ersatz system to send you the 7zads file.
        
        Best regards
        François
        
        -
        François Boone, ing., Ph. D.
        
        Professeur,
        Directeur du département de génie électrique et de génie informatique
        Université de Sherbrooke
        2500, boul. de l'Université
        
        Tél. : 819 821 8000 poste 65900
        Fax : 819 821 7937
        Courriel : infogegi@usherbrooke.ca
        www.usherbrooke.ca/genie
        
        Le 21 août 2018 à 02:16, volker_muehlhaus <moderator_forums@keysight.com> a écrit :
        
        Keysight Community
        Re: Momentum vs FEM: what's the problem ?
        reply from volker_muehlhaus in EDA Software - View the full discussion
        Hello François,
        
        you have one edge where both signal and ground line end, and it seems that ADS uses the entire edge length of the ground plane for the port reference width (instead of just a part of the edge underneath the signal line). In Momentum that causes trouble, in FEM I don't know. Can you create an archived ADS project with this antenna and upload it here?
        
        Best regards
        Volker
        Reply to this message by replying to this email, or go to the message on Keysight Community
        Start a new discussion in EDA Software by email or at Keysight Community
        Following Re: Momentum vs FEM: what's the problem ? in these streams: Inbox
        This email was sent by Keysight Community because you are a registered user.
        You may unsubscribe instantly from Keysight Community, or adjust email frequency in your email preferences
        >
        Like • Show 0 Likes0
        Actions
        volker_muehlhaus @ Fresnel on Aug 21, 2018 11:00 AM
        Mark Correct
        Correct Answer
        Hello Francois,
        
        thanks, I received your workspace and made some changes to the models.
        - For Momentum, in emSetup I changed the physical model to 3D, so that it uses thick metal modelling instead of infinitely thin sheets.
        - I changed the ground pin to an "edge/area pin" to make sure it doesn't extend across the entire edge.
        - I didn't use any port calibration, because I don't trust that for this explicit ground case (my personal view, might be wrong).
        - For FEM settings, I started from scratch. Frequency for mesh = 2GHz (instead of fmax), boundary distance 40mm.
        
        I get rather similar results from Momentum and FEM now, which are somewhat different from yours.
        There is one warning on my port in FEM that I don't understand, maybe that is something for Keysight support.
        
        The packed project with my models is here:
        http://muehlhaus.com/wp-content/uploads/2018/08/Type_6_volker_wrk.zip
        
        Best regards
        Volker
        
        1 person found this helpful
        Like • Show 0 Likes0
        Actions
        Fresnel @ volker_muehlhaus on Aug 21, 2018 1:11 PM
        Mark Correct
        Correct Answer
        Thank you very much for your reply.
        I will study carefully your projet to be able to obtain same results as you: a quick test gives similar results but with a difference about 1 dB compare to yours.
        
        Best regards.
        François
        
        -
        François Boone, ing., Ph. D.
        
        Professeur,
        Directeur du département de génie électrique et de génie informatique
        Université de Sherbrooke
        2500, boul. de l'Université
        
        Tél. : 819 821 8000 poste 65900
        Fax : 819 821 7937
        Courriel : infogegi@usherbrooke.ca
        www.usherbrooke.ca/genie
        
        Le 21 août 2018 à 14:01, volker_muehlhaus <moderator_forums@keysight.com> a écrit :
        
        Keysight Community
        Re: Momentum vs FEM: what's the problem ?
        reply from volker_muehlhaus in EDA Software - View the full discussion
        Hello Francois,
        
        thanks, I received your workspace and made some changes to the models.
        - For Momentum, in emSetup I changed the physical model to 3D, so that it uses thick metal modelling instead of infinitely thin sheets.
        - I changed the ground pin to an "edge/area pin" to make sure it doesn't extend across the entire edge.
        - I didn't use any port calibration, because I don't trust that for this explicit ground case (my personal view, might be wrong).
        - For FEM settings, I started from scratch. Frequency for mesh = 2GHz (instead of fmax), boundary distance 40mm.
        
        I get rather similar results from Momentum and FEM now, which are somewhat different from yours.
        There is one warning on my port in FEM that I don't understand, maybe that is something for Keysight support.
        
        The packed project with my models is here:
        http://muehlhaus.com/wp-content/uploads/2018/08/Type_6_volker_wrk.zip
        
        Best regards
        Volker
        
        https://community.keysight.com/servlet/JiveServlet/downloadImage/2-87132-157856/mom_fem.PNG
        Reply to this message by replying to this email, or go to the message on Keysight Community
        Start a new discussion in EDA Software by email or at Keysight Community
        Following Re: Momentum vs FEM: what's the problem ? in these streams: Inbox
        This email was sent by Keysight Community because you are a registered user.
        You may unsubscribe instantly from Keysight Community, or adjust email frequency in your email preferences
        >
        Like • Show 0 Likes0
        Actions
        Riju @ volker_muehlhaus on Sep 19, 2018 2:04 PM
        Mark Correct
        Correct Answer
        The workspace you created is very helpful.
        I have generated the 3D view of your FEM design and found that due to the internal area port, the impedance line is slightly slanted like the following picture.
        It is generating the following warning
        --- WARNING -------------------------------------------------------------------
        
        Port P1 has one pin with a segment and one pin without a segment. This may result in a triangular sheet port. Use an Edge/Area pin to associate a line segment with the negative pin to adjust the shape of the port.
        
        -------------------------------------------------------------------------------
        
        =============================================================
        I changed the area ground pin to edge pin where the edge length is equal to the width of the top plane microstrip (here it is 3.18 mm) and the edge pin is connected exactly in the same X-Y location/co-ordinate of the + pin. This removes the triangular sheet port warning.
        If I use uncalibrated (Direct) edge port I get the rectangular shaped port like the following
        
        Now, for TML case, the waveport extends below the ground plane like following. Is the shape of the waveport correct or it needs to be re-sized? How can it be re-sized so that it can't go below the ground plane?
        I observed dissimilarity in the results between this extended calibrated TML waveport and uncalibrated Direct port.
        Like • Show 0 Likes0
        Actions

Momentum vs FEM: what's the problem ?

Attachments

Outcomes

Attachments

Related Content

Recommended Content