Jump to content

JIF

Moderators
  • Content Count

    403
  • Joined

  • Last visited

  • Days Won

    21

Everything posted by JIF

  1. Hello Sourabh13, Yes, your understanding is correct.
  2. JIF

    Error In Feko

    Hello Sra, The windscreen method was developed to model antennas embedded in windscreens and one of the assumptions are that the windscreen is relatively flat, but still curved. This error is given when two triangles in a windscreen reference face have normals that are separated by more than 90 degrees and thus probably have too much curvature for the assumptions used in the windscreen solution method. Without the model, I'm not able to confirm this or comment further on the applicability of the windscreen solution method to your model.
  3. Hello shashank119, I understand the request, but this is not currently possible in FEKO, but has been requested and should be available in the future.
  4. JIF

    Import ERROR Feko

    Hello Pawel, The student edition of FEKO does not support geometry import or export - see https://altairuniversity.com/feko-student-edition/. You can import and export Parasolid, since that is the native format used in FEKO. If you can export a Parasolid file, you will be able to import it into FEKO.
  5. Hi ben, You are correct; when triangles intersect or don't align where they touch, FEKO will give an error. FEKO 2018 has improved mesh checking in the solver and thus it is more likely to detect the intersecting / overlapping elements. You need to correct your model to ensure valid mesh connections. This is required in the MoM since basis functions need to created over triangles and this requires a valid mesh.
  6. I suspect it should be: !"C:\Program Files\Altair\2017\feko\bin\cadfeko.exe" --run-script BuildModelAndSimulate.lua --non-interactive Note that I have put quotes around the application and in your prefeko example and the rest is passed as a parameter as that seems to be what you did for "filename.pre".
  7. I already provided the details in my previous comment. If you don't need to change the name of the stl or cfx file, then the command (in a FEKO terminal) would simply be: cadfeko --run-script BuildModelAndSimulate.lua --non-interactive You should be able to run CADFEKO (or the command above) from Matlab, but then you will have to use the full absolute path to the cadfeko application binary (since the environment won't be setup inside Matlab). I don't have access to Matlab, but I know many users have called FEKO components from external applications, including Matlab. I hope this answers your question. For extracting the data, you may want to consider a Lua script for POSTFEKO to write the data that you want to disk. See
  8. What you are trying to do is very easily possible, but there are many ways to do the same thing and I'll pick one (maybe not the one that works best for you). The Lua scripts are CADFEKO (or POSTFEKO) scripts and thus you need to use them in CADFEKO (or POSTFEKO), not runfeko. The runfeko application will take a FEKO model (either a *.cfx file, *.pre file or *.fek file) and performs the simulation. FEKO only understand *.fek files, but runfeko knows what to do with *.cfx and *.pre files to create the *.fek file and then run FEKO (that produces the *.bof, *.out and many other files). I looked at your recording and most of it is fine, but there are a few problems and the crucial bit is missing. Firstly, since you are creating the mesh in an external application and then importing it, there is no reason to remesh in CADFEKO, so I removed that section. The crucial bit that you missed is that you also need to define the layered dielectric (in your case, just a single 0.6 um layer of Graphene) and then apply it to the imported face. I have adjusted the script accordingly (this is similar to what I did previously with your *.pre file). You have a few options, you could either create a script that creates the and saves the CADFEKO model (*.cfx) and then run runfeko on that model or (as you did), launch runfeko from withing CADFEKO (in the script). I would probably rather split the two (create CADFEKO model and run the solver), but for now I decided to just stick with your script and run it from CAFEKO. Here is the modified script: BuildModelAndSimulate.lua. As I have mentioned, you need to run the script in CADFEKO, but that does not mean that you have to click it manually. You can run CADFEKO in non-interactive mode from command line or an external application. Here is an example: cadfeko --run-script BuildModelAndSimulate.lua --non-interactive --configure-script "modelName='testName.cfx' ; stlName='graph_001.stl'" | more As you can see, it runs CADFEKO, provides the script that CADFEKO should execute, ensures that the GUI does not display (if you remove this, CADFEKO will remain open after the script has completed, but with this option it will close once completed) and then it also passes some configuration (just normal Lua code that is executed before the script execution starts) and finally the output (stdout) is piped to more (I added this so that I can see errors in the script since GUI applications don't write to the console in Windows - in Linux this is not required). As part of the script, the solver will also be run, but as I mentioned, if I had to do this, I would remove that from the script and do a second call to runfeko instead (I'll leave that simple change to you). With the steps above, you can adjust the name of the stl file to import and the name of the resulting model and thus it easily fits into any automated system. For reference, here are the results (0.6 um Graphene compared to PEC): It is not the same as your graph, but after changing the *.pre file to also use 0.6 um Graphene, the results match perfectly (as expected). Yes, 0.001 was the thickness. If you change it to 0.6, you get the results above? I don't want to explain each parameter, since it is not required. When you are on a card in EDITFEKO, press F1 and then modify the parameters in the panel. The same goes for the 4 and the -1 of the SK card - don't worry about them and use the panel and they will be adjusted to the correct values corresponding to the options selected and entered in the panel.
  9. Hello Khadimul, My first question when EDITFEKO is being used is always, why? Everything that you want to do (what you have tried to do thus far) is possible in CADFEKO, also using scripting.Consider using CADFEKO, but there is nothing wrong with using EDITFEKO, so I'll respond to your EDITFEKO question. The *.pre file that you have defines a dielectric, but it never applies that dielectric. You need to apply a medium with the ME card (for dielectric volumes) or with the SK card (lossy metals and thin dielectric sheets). Your model is a simple plate and not a closed volume. That means that you can't model it as a dielectric volume (a medium on the one side and a different medium on the other side). Since I don't have control over the mesh, I decided to model it as a thin dielectric sheet, but then I need to define the sheet thickness and I picked an arbitrary value that you will have to modify. The layered dielectric is defined using the DL card and uses the definition created with the DI card. Finally, the thin dielectric sheet is applied using the SK card. I have included the modified *.pre file as reference. Obviously it needs to be modified by you, but it shows how to do it. DONE_mod.pre
  10. Hello shan25, I don't have much experience with metamaterials, but FEKO does support them. However, for a gradient index media, the media properties need to change as a function of its position (correct?) and that is not supported in FEKO at the moment. FEKO only supports media properties that vary as a function of frequency.
  11. Hello Kalyan, No, my comment was w.r.t. the finite thinkness of the track, not the substrate. For MoM simulations, the tracks are usually not modeled as finite thickness strips. That doesn't sound like a good idea. I think my comments may have caused more confusion. You can model the substrate as a finite substrate using SEP (default MoM dielectric solution) or as infinite (using planar multilayer substrate). As I have mentioned, my comment was w.r.t. the microstrip tracks and not the substrate. My comments were probably influenced by what I noticed in the model added to:
  12. Hello Kalyan, I started running A1.cfx and I have already passed the point where you ran into an error. However, there are some differences - I ran on Windows and ran with the latest update to 2017.2. Thus, we are not running the same version and not on the same OS. There was a bug fix in one of the updates of 2017.2 that resolved (workaround for) a problem with the Intel compiler and some (newer) Linux versions. Are you able to update to the latest version via the updater?
  13. Hello Kalyan, The problem seems to be with the parallel eigen solver. I ran the model for a few minutes without issues when running sequentially, but when running in parallel, I get: Characteristic mode analysis of MoM matrix equation Extracting real and imag. parts of the MoM matrix eqn. Calculating the eigencurrents using ARPACK The following message from the server process (MYID= 1): The following message from the master process (MYID= 0): ERROR 40161: An error occurred while initialising the eigensolver I'll log an issue so that the developers can investigate the reason for the problem. For now the workaround is to run sequentially. As a side note, I would remove the finite thickness in the model (usually not the best way to create a MoM model, unless it is critical in the particular application).
  14. Hello Kalyan Durbhakula, I can't help you or look at the DXF import, since you only added the imported model (result), but I can help you with the alignment to fix the problem. CADFEKO has a number of alignment tools including a translation and a rotation tool. For your model it looks like a simple translation would be sufficient. The more general alignment tool allows the user to align a source workplane with a target workplane as in the recording below: For your specific case, the more simple "translate" tool should be sufficient. See below. Unless this is a very thick track, the thickness is usually not modelled in MoM by creating a finite thickness. I would rather not do the sweep, unless it is really an issue for your model and needs to be taken into account (it will also require a much finer mesh).
  15. If you are trying to model an existing system, I would suggest that you take a closer look at the antennas on the tower. I'm pretty sure that they are not monopols (even if they look like monopoles). They are most likely colinear antennas. The single ground is not the problem - you can just add the ground under the tower to take the effect of the ground / earth into account.
  16. Hello Khadimul, What was the error? Elements too large? The element sizes are specified at the IP card and it looks like you set it to 3.7. It also looks like you are using a scaling factor (SF card) and all dimensions are in microns (scale factor = 1e-6). The PREFEKO language also supports the RM card if you really need to remesh some of the face labels, but since you are importing everything from one file, the IP card should do what you want. Note that CADFEKO also supports automation via scripting. It should be very easy to do in a script what you can do manually in CADFEKO. You can even record the steps (macro recording) and then just modify the script after the recording as you need. I hope this helps. If not, please add more details inculding the STL file so that I can take a look.
  17. Hello Khadimul, There are various ways to start CADFEKO. Assuming you know where FEKO was installed, you can run FEKO in one of the following ways: Launch the application that you want to start using the full path to the application. As an example, if FEKO was installed in "/opt/feko/2017", run "/opt/feko/2017/altair/feko/bin/cadfeko". Launch it from the HyperWorks scripts folder. As an example, if FEKO was installed in "/opt/feko/2017", run "/opt/feko/2017/altair/scripts/cadfeko". Source initfeko and then simply call the applications directly. As and example, if FEKO was installed in "/opt/feko/2017", run ". /opt/feko/2017/altair/feko/bin/initfeko" and then simply run "cadfeko". In this example, note that sourcing initfeko starts with a dot, followed by a space and then the path to initfeko. In all examples above, don't include the quotes. If you still have issues, please provide me with your Linux distro details. I know Ubuntu uses dash instead of bash and that could cause issues (I usually go back to bash instead of dash).
  18. Sorry mustaphos, I have been trying to answer you all day - other things keeping me busy (also traveling). I think we need to step back and define what you are trying to model. Are you trying to model an existing system or are you designing a system? If you are designing a system, I think you should reconsider your antenna choice. A monopole is not the antenna that you should use here (probably). For a monopole, you are going to have to add a ground plane in some form or another. Anything other than a perfect PEC ground plane is an approximation to what you want. Options include the ground planes that you have added or an approximation to a ground plane using wires or something similar. You could also use your structure (mast) as a ground plane, but I would not recommend this - it is going to severely affect your antenna performance as you modify the structure and connection to the structure and is not a practical solution. If you are simply trying to create a model of something that already exists, I would take a closer look at the antenna. The omni antennas are probably not simple monopoles, but probably a collinear antenna array.
  19. Everything that you want to do should be possible. You would probably want to model the ground (earth) using a real ground (not PEC). You can take a look at "Yagi-Uda Antenna Above a Real Ground" in the Example Guide as an example to add a real ground. If you have a PEC ground, then you can model that as well. Or you can model it with a finite metal ground (much more expensive in terms of simulation resources). The features in FEKO are designed around what is possible in reality. It thus helps to think about what happens in reality when creating the model. If you have a monopole antenna, it will surely assume to be attached to a ground plane, even if that ground plane is the pole that it is mounted on. If that is the case, add your excitation between the antenna and the ground structure. You can also create a more realistic model by simply feeding the antenna with a transmission line - you would then be modelling the cable feeding the antenna. If you need more help, I suggest that you add the model or at least some pictures.
  20. Hello CrisWu, Are you asking if it is possible to run FEKO Lua scripts in CADFEKO or POSTFEKO? CADFEKO and POSTFEKO have options to allow running scripts, configuring scripts (init variables) and run in non-interactive mode (no GUI). These are described in the documentation, but I have repeated them below. --non-interactive Special execution mode for running automation scripts without user interaction. --run-script SCRIPTFILE Specifies an automation script to load and run. --configure-script CONFIGSTRING Executes the string CONFIGSTRING before running the script specified in SCRIPTFILE. This options is only used with the “non-interactive” option.
  21. Hello mustaphos, Just create a ground plane and add multiple monopole antennas (port and excitation for each antenna) and you have multiple monopole antennas with a single ground. The ground can be finite or infinite (planar multi-layer substrate). Or did I misunderstand what you are trying to ask?
  22. Hello, I looked at your model and the first thing that I noticed is that the meshing was too coarse. You have selected "Fine" meshing and usually that would be sufficient for a fine mesh, but in this case geometry (line widths and even lengths) are very small compared to a wavelength. Are you sure about the material parameters, dimensions and inductance value? When I simulate the model (with a refined mesh), the resonance is around 3.5 GHz - thus a factor 4 higher than where you expect it. Could you please verify the parameters and dimensions. Maybe a photo of what you measured so that we can see what you built matches with what you are trying to model? An example of a reasonable (maybe a bit fine) mesh for this model looks like this: What I have done is to ensure that the triangles representing the line is not wider than the line itself. I have also ensured that the triangles are are not larger than the height of the substrate. These are general rules that I use to guide meshing.
  23. Hello manuel_lulea, Thanks for the image - it makes it much easier to see what you are doing and provide meaningful suggestions. To get back to your original question of " Is it mandatory to have a ground plane for the simulation?": FEKO does not require a ground plane - you can simulate what you want, but that does not mean that the antenna will work the way you think it should. For your meander line model, you need to add a ground plane for the antenna to work. You can think of the meander line antenna as a monopole - a monopole won't work without some sort of ground plane since it is using the ground and its electric symmetry to model one half of a dipole. I hope this makes sense to you. Maybe this helps: Start with a dipole, then turn it into a monopole by halving it and adding a ground plane and then compact the monopole into a meander so that it takes up less space. There is nothing wrong, from FEKO's perspective with what you are doing. Instead of wires, you can use edge ports, but they have limitations on the boundaries of SEP dielectrics (does not look like you are using SEP though). Have a look at the article below if you do and want to use edge ports with SEP dielectrics since it shows how they should be constructed. For your antenna to work, you will have to add a ground plane. If this is a microstrip feed, the ground plane should be added at the bottom, but not under the antenna - it should be added to the section of the model that currently has nothing (on the side of the voltage source). I found an image that shows what you should be constructing (see below). What you are missing is the bottom section in the image (note that there is a ground plane on the bottom, but only on a section, not under the antenna). If you don't understand, send me your model (not just a picture) and I'll modify it for you.
  24. Hello mustaphos, Am I correct when I say that you don't get this error (hydra error) for every run, but it happens sporadically? We have seen this problem a few times and it is a bug in Intel MPI. It has been reported to them and we are looking at newer versions to see if the problem has been resolved, but thus far they have not been able to resolve the problem. Running the simulation a second time should not result in the same problem (that has been our experience thus far).
  25. JIF

    Meshing problem

    Hello Vikas88, I suspect that you are simply running into issues due to the large number of elements, but even if you do mesh the model (and we could provide some suggestions if you provide a better problem description), how are you going to solve it (you will need a very large machine or cluster)? Is the solution not rather to use some of the other features or solution methods in FEKO? If you give us a description of the problem, we can suggest alternatives that reduce the number of required elements (HOBF for MoM or MLFMM if you need a full wave solution or PO, LE-PO, RL-GO, UTD for an asymptotic solution).
×
×
  • Create New...