Jump to content

andries

Members
  • Content Count

    47
  • Joined

  • Last visited

  • Days Won

    2

andries last won the day on November 9 2017

andries had the most liked content!

About andries

  • Rank
    Advanced Member

Profile Information

  • Gender
    Male
  • Are you University user?
    No

Recent Profile Visitors

1482 profile views
  1. hi, my model was meshed fine when it was just PEC,i coated it with a dielectric material and it is now showing error 2331.kindly help

     

  2. Hi @mahdi72, Lightning analysis and cable modelling are both quite complex fields of study. This means there is no quick answer to your question, although you might want to start with: "Lightning" related topics: http://www.altairhyperworks.com/ResourceLibrary.aspx?keywords=lightning&altair_products=FEKO "Cable Harness EMC/EMI: Cross Talk, Radiation, Interference & Susceptibility": http://www.altairhyperworks.com/ResourceLibrary.aspx?keywords=Cable+Harness+EMC%2fEMI%3a+Cross+Talk%2c+Radiation%2c+Interference+%26+Susceptibility&altair_products=FEKO
  3. Great! I'm glad it helps. It's definitely worth it to go through the Examples Guide example, though. It takes you through the basic concepts and shows you how to navigate the API documentation.
  4. Hello @mustaphos, @JIF beat me to it...! Of course, this is the "more correct" way to learn how to do scripting and automation. The shortcut way to get started with scripting (in CADFEKO) is to use "Macro recording". You'll see in the status bar (in the bottom of the application) a button with a puzzle piece and a red circle. If you press this, then CADFEKO will record every action you take that changes the model. In other words, if you start recording and create / modify / delete a variable, then a script will be recorded that shows you how to do precisely that one bit of scripting. It's cheating but I won't tell anyone.
  5. Hello! The trigonometric functions take radians; did you convert the angles before using them? In other words, multiply the angles by pi/180 inside the cos and sin functions. Let me know if that helps!
  6. Hi @Waqas Hayat, I'm not at a computer where I can open the model but it sounds like a question answered in This explains what EPSENT is and what could have caused the error. Other tools that you can try include: Using some of the CAD fixing tools (usually helpful if you have an imported model); or Changing the model's bounding box size (this influences what the mesher sees as "the same" point and might help to avoid an error) I can try to open the model tomorrow but in the meantime maybe you can try a few things as well and let the forum know if anything (and what) helped!
  7. You can maybe look at the Examples Guide. Example D-1 is "Shielding factor of a sphere with finite conductivity" which is conceptually similar. There is more information on the formulations in the User Manual (look for the description of the "SK card") but this might be a bit too much detail for your purposes.
  8. Hi @MOHAMED HAMDALLA, At higher frequencies, the skin depth of a metallic region quickly limits how deeply electromagnetic waves can penetrate a 3D object. In FEKO, then, the method to model a metal is to: set the inside region to "free space"; and to set the face properties to a metal with a given thickness See evo2_modified.cfx for an example. Note that I've also change your frequency settings to a single 1GHz frequency point. You can, of course, change it back to a frequency range but I wouldn't expect the results to make sense all the way down to 1Hz. The geometry is in the order of 22cm long, which is electrically a very small object at lower frequencies.
  9. Hi Roberto, Can you confirm which version you are using? There was a similar problem with WinProp 14 but the WinProp 14.1 update provided a fix (released end of June). If you're not using 14.1 or newer, then you can try to update to the latest version.
  10. Thanks @ALM, I'm attaching three models that you can take a look at. editfeko_forloop.pre This is essentially what you provided above and (as far as I can tell), it works as expected. The PREFEKO language (what you use in EDITFEKO) can be quite tricky. New solution configurations are triggered by certain cards (e.g. a new source and a solution request). If your loop didn't contain certain triggers, then it's possible you wouldn't have seen all ten configurations. The model that I attached shows that your loop is configured correctly, provided the rest of your configuration is fully configured. The downside to this method is that it is more complex to configure and results in ten independent configurations. editfeko_freqrange.pre This model shows a simpler method to request a frequency range in EDITFEKO. The FR card has an option for you to specify a start frequency, end frequency and the number of points: FR: #freq_Samples : 0 : : : : #start_frequency : : #end_frequency FEKO will then take care of the loops and store the results in one configuration. cadfeko_freqrange.cfx Finally, I'm adding a CADFEKO model (created with CADFEKO 14.0.432-293039 (x64) from 2016-12-06) that requests a frequency range from 2.2GHz to 2.5GHz using 10 samples. This is by far the simplest (and safest) way to configure a model and would be the recommended method. My best guess is that you were using the "Continuous (interpolated) range" option to specify the frequency. For this setting, FEKO will decide for itself how many points are needed in order to accurately represent the simulation data. In your case, it's possible that the algorithm was satisfied to stop after four points. If the results are inaccurate, you can increase the "convergence accuracy" to "High". Otherwise, your problem could be that the mesh is too coarse or there is some other problem with the model setup. If this still doesn't help, please attach an example model so that we can see what you're working with.
  11. Hi @ALM, Could you attach a model? Did you set up the configuration in CADFEKO or in EDITFEKO (I'm asking because you mentioned a for loop)? You can maybe check whether the maximum number of samples or the minimum step size is set. These both serve to limit the number of samples and could result in something like this. Another problem could be that the convergence algorithm isn't checking whether all of the results are converged. The defaults are usually fine but if you are looking at surface currents, then it might be switched off. Alternatively, you can try using linearly spaced points for now. I hope this helps!
  12. Hi Mohamed, I don't think I have a compatible version anymore. If you run straight wire example.lua though, then the script should rebuild the model from scratch. To run the script, just drag and drop it into CADFEKO and click on the run button.
  13. Hi Mohamed, Thanks for the model. There are a few things that could explain why you're not seeing the characteristic modes in the attached model: The model doesn't contain any geometry. There is only an infinite ground plane and a cable instance, both of which are approximations that can be used instead of complex geometry. For characteristic modes, a model that will be solved using the Method of Moments (MoM) is needed. In other words, there typically needs to be CAD that can be meshed into triangles and wire segments. The ports for the model are both Finite Element Method (FEM) line ports. FEM is a separate solver from MoM and doesn't support characteristic mode analyses. Neither port is excited or connected to anything. I recreated what I think the model is trying to do (see stright wire modified.cfx) using MoM. Note that you need to add the requests that you are interested in; at the moment only the characteristic mode eigenvalue data is being calculated. Are you trying to reproduce results from an article/measurements/calculations? Do you have any reference data? Without further information, this is as far as I can go. I would urge you to go through the Examples Guide and some overview documents on the website: https://altairhyperworks.com/ResourceLibrary.aspx?keywords=Characteristic&altair_products=FEKO
  14. Hi Mohammed, Can you attach the model? It will make it a lot easier to see what's happening. It sounds like you don't have a valid mesh for the dipole. How long is the dipole and at what frequency are you performing the simulation? What is the wire radius? You can also compare your model to the Examples Guide model ("A-1 Dipole example") and see if anything stands out.
  15. Hi @bcoles, I didn't look at your model in detail but can give a few general comments: At your design frequency, the free space wavelength is about 4.7m. The coils in your model have dimensions <20cm, which makes them electrically quite small. In these cases, the default mesh sizes might not be sufficient and you'll need to do a mesh convergence study. Maybe have a look at error estimates and adaptive mesh refinement. A final note on meshes is that ports are typically quite sensitive; make extra sure that those are fine enough. When you are looking at S-parameters (here e.g. S12), you are looking at "how much of the power given to port 1 is reaching port 2". This implies that some of the power can get diverted elsewhere. With a well-matched input port (as in your case) we know that power lost due to reflections is quite small. However, you still need to consider the power that is dissipated by material losses, power absorbed by other ports or power that gets radiated away. Basically, once you know that you can trust your mesh, try to find all of the power sinks in your model. This won't necessarily improve S12 but at least you'll know where the power is going. Regards, Andries
×
×
  • Create New...