Release Notes: Altair Feko 2019.2.1

Altair Feko 2019.2.1 is available with new features, corrections and improvements. This version (2019.2.1) is a patch release that should be applied to an existing 2019 installation.

Feko is a powerful and comprehensive 3D simulation package intended for the analysis of a wide range of electromagnetic radiation and scattering problems. Applications include antenna design, antenna placement, microstrip antennas and circuits, dielectric media, scattering analysis, electromagnetic compatibility studies including cable harness modelling and many more.

WinProp is the most complete suite of tools in the domain of wireless propagation and radio network planning. With applications ranging from satellite to terrestrial, from rural via urban to indoor radio links, WinProp’s innovative wave propagation models combine accuracy with short computation times.

Feko 2019.2.1 Release Notes

The most notable extensions and improvements to Feko are listed by component.

CADFEKO

Features

  • Added API support for importing cable data from a .kbl file.
  • Extended export to I-DEAS universal format (.unv file) to include media information.
  • Changed the default format on the Import near field dialog from Sigrity to Feko Solver field on Cartesian boundary. Please modify Lua scripts that rely on the default by explicitly setting the DataType property of the NearFieldDataFullImport object.

Resolved Issues

  • Resolved a crash that could be encountered when running a Lua script that makes use of form dialogs to gather user input.
  • Added validation to prevent modifying a medium that is used by a locked mesh part. This caused an assertion failing with the message src\mediaframework\gaia_MediaLibrary.cpp (751): Assertion failed: 0.
  • Fixed a bug where, when running the command-line mesher (CADFEKO_BATCH) on a machine with locale settings that use a comma (,) as the decimal separator, the setting for using higher order basis functions was ignored. This resulted in a significantly larger number of mesh elements than expected.
  • Fixed a regression that got introduced in CADFEKO 2018.1 that made the radius of model mesh segments non-modifiable.
  • Fixed a problem with the renaming of cable elements through automation where the labels in the tree and the 3D view did not update correctly.
  • Resolved a problem with the updating of a cable harness label following a rename. The label from before the rename was written to the .pre file for the cable cross section definition(s). Instances of the old label and of the new label could be observed in the POSTFEKO model browser. The updated label was correctly written out after the model was closed, re-opened and saved. The correct labels are now used consistently.
  • Resolved an issue in the CADFEKO API with the recently introduced ClosestVertexTo method. This method can now be used for the CurvilinearSegments and CurvilinearTriangles objects to provide the vertex nearest to a specified point.

EDITFEKO

Resolved Issues

  • Corrected the FF card to recognise the value when selecting the option to Calculate only the scattered part of the field for far fields using Cartesian coordinates.
  • Corrected the CR card labels for Position (coordinate) and Rotation about the axes. The position and rotation are specified in Cartesian coordinates, regardless of the coordinate system setting. The coordinate system determines how the tensor is interpreted.

POSTFEKO

Features

  • Changed the default Quantity settings (for example, dB and Normalise) in the Result palette when adding a result to a graph or 3D view that already contains a result of the same type. POSTFEKO will apply the Quantity settings from the last request of the same result type to the newly added result. There is no change to the default Quantity settings for the first result that is added to a graph or view.

    For example, if a Cartesian graph already contains an S-parameter result in dB, the next S-parameter result added to the Cartesian graph will also be plotted in dB.

    In cases where existing Lua scripts used the default settings, executing the script may fail, or graphs and 3D views could display the wrong quantities. The scripts should be modified to explicitly set the required quantities.

  • Extended the API with the DRE namespace that provides an interface to import/export results from HDF5 files in the DRE format.

Resolved Issues

  • Fixed a crash that could be encountered when viewing optimisation results plotted on a Cartesian graph while running the optimiser.
  • Resolved a crash that could be encountered when restoring the POSTFEKO window after executing a script that plots results while the application is minimised.
  • Resolved a crash that could be encountered when running a Lua script that makes use of form dialogs to gather user input.
  • Resolved an issue with bi-static RCS values that were plotted incorrectly when the source used the setting Calculate orthogonal polarisations and the .fek file was present.
  • Resolved an issue where the slicing panel did not include the Polarisation angle when multi-selecting far field and stored far field traces.
  • Resolved an issue where setting the animation type was not stored correctly and would reset to the first item in the drop-down list. For example, setting the animation type to Phase and changing the selection or selecting to export the animation would change the animation type to Frequency.
  • Resolved a performance problem when selecting a result in the tree belonging to a model that contains multiple looped plane wave sources or a characteristic mode analysis request. The previous slow performance was caused by validation checks that are carried out to determine whether the results are valid for export. After the change, the slow behaviour will only be experienced when selecting to export a result from such a model. The results can now be plotted without delay.
  • Resolved an assertion that failed with a message referring to common_AxisSet.h when exporting a custom dataset containing an axis with string values.
  • Fixed an assertion that could fail in POSTFEKO 2019.2 with the message Assertion failed: delta.isEmpty() when interacting with a 2D graph from a .pfs file saved in an older version.
  • Resolved an assertion that failed when adding electric or magnetic scalar potential near fields (the scalar values, not the gradients) to a graph with only the .bof file present.
  • Resolved an assertion that failed with message ending in common_MultiAxisSet.cpp (89): Assertion failed: isSingular(). This assertion failure could be encountered when opening a model containing multiple near field requests with the same label (in different configurations) where at least one of the requests was a Cartesian boundary near field with multiple surfaces.
  • Fixed the API DataSetAvailable property on the NearFieldData object to correctly return false for a Cartesian boundary request with multiple surfaces. This request type is not currently supported in the API.
  • Added validation that prevents getting the data set of a Cartesian boundary near field with multiple surfaces. The API is not yet extended with support for Cartesian boundary near field data sets. An assertion could fail in POSTFEKO 2019.2 with the message Assertion failed: markedMultiAxisSet.hasSingleMarkedAxisSet() if the GetDataSet method was used on a Cartesian boundary near field result.
  • Resolved an issue where using StoreData to save a result as a different dataset type resulted in the dataset being stored despite an error being given. For example, storing a near field dataset as a far field would incorrectly store the data as a far field stored data result.
  • Updated the parameter sweep script to ignore results that are not available. This could be due to simulations that aborted or results that did not converge. A message is displayed with a list of the runs that are excluded from the combined results.
  • Updated the Parameter sweep: Combine results application macro to support Cartesian boundary near field results with a single surface. The script ignores merging data for Cartesian boundary near fields with multiple surfaces or any other faulty requests.

Solver

Resolved Issues

  • Improved the robustness of ray/geometry intersection tests, during an RL-GO solution, to prevent corner cases where rays penetrated closed conducting surfaces.
  • Fixed a bug that triggered a floating point exception when computing the transfer impedance of Demoulin/Tyni shield with minimum optical coverage set to 100%.
  • Added support for current sources on FEM line ports within an anisotropic FEM region.
  • Improved the time efficiency of the computation of interactions in free space during the right hand side calculation phase of the solution for models with impressed current sources.
  • Refined the accuracy settings for calculating the per-unit-length parameters of unshielded cable bundles.
  • A short-circuit connection is now allowed between a subset of pins of a black box circuit or a general non-radiating network that has more than two exposed pins, that is connected to a cable.
  • Fixed a bug that resulted in large reflection coefficients being computed for models with lossy multilayer substrates solved with RL-GO.
  • Improved the accuracy of source power calculations for models with impressed near field sources.
  • A warning is now issued when compression of plane waves, in a loop with multiple directions of incidence, is not applicable and a phase corrected initial guess is applied during the iterative solution when reverting to the solution with each plane wave in the loop as excitation.
  • Fixed a bug that led to an error state when solving a combined MoM/MTL cable harness with a double shield.
  • Fixed a bug that could have resulted in a message passing interface (MPI) buffer overflow.

Shared Interface Changes

Features

  • Added an additional POSTFEKO extraction script to the Feko-HyperStudy interface. If a POSTFEKO session is available in the study folder (without an extraction script), the new extraction script is created in the folder when importing variables in HyperStudy. The script extracts all the visible traces on the Cartesian and polar graphs in the session to the HyperStudy output file. The HyperStudy output file can then be accessed in HyperStudy to set up an output response to be used in a DOE or optimisation run.
  • Added a tree widget for Lua forms. The FormTree object is available in the CADFEKO and POSTFEKO API.

Resolved Issue

  • Resolved an issue with the HyperStudy-Feko utility where the file dependency list was not populated correctly.

Support Components

Features

  • Reduced the licence checkout times for some components.
  • Updated the response surface-based optimiser to the latest version.

Resolved Issue

  • Fixed a bug that led to a crash in OPTFEKO on Windows.

WinProp 2019.2.1 Release Notes

The most notable extensions and improvements to WinProp are listed by component.

General

Feature

  • Changes made to the .net file during optimisation with OptMan are now saved automatically. This avoids pop-up windows at every iteration.

Resolved Issue

  • Improved error messages and warnings in OptMan. When an OptMan simulation is launched without an optimisation target, the error message will state what is missing. Further, some unnecessary warning messages were removed.

ProMan

Features

  • The delay of the shortest path from transmitter to receiving point can now be computed and is available for display in the results tree.
  • Added support for the 512-QAM and 1024-QAM modulation types.
  • Improved the performance of the ProMan GUI when dealing with large databases. The response time of the GUI to various click events, as well as the loading time for displaying results in the 3D view, are reduced.
  • Added support for the definition of a rotation angle that can be applied to rotate all receiving array elements in the azimuth plane, for the case when receiving antenna is defined using the Individual location offset for each element option.

Resolved Issues

  • Fixed a bug that resulted in the effects of the direct ray not being taken into account during simulations with the intelligent ray tracing (IRT) model in urban scenarios.
  • Fixed an issue where displaying a result for time-variant simulations in point mode, did not show colour in the result pixel.
  • Fixed a bug that led to a corrupted visualisation of results in ProMan for a rural project with topography defined in geodesic coordinates.
  • Fixed a bug in the interpolation of topographical elevations. This improved the accuracy of results in rural scenarios. Also, the topographical height at the location of the transmitter is now correctly computed.
  • Improved the accuracy of predictions inside buildings in an urban scenario.
  • Fixed a crash that could occur in a rural propagation simulation when Channel Impulse Response was requested as output while Propagation Paths were not requested.
  • The east over north angular convention is now used in the spatial channel impulse response and angular profiles (MS, BTS).
  • Fixed a bug that resulted in the wrong longitude being written to the .kml file when exporting prediction results, as a bitmap, to Google Earth.
  • Improved the wedge detection algorithm to correctly handle wedges between a wall with disabled diffractions and one with enabled diffractions.
  • The computation of angular means at the base and mobile stations is reactivated as a result type obtainable from propagation analysis.
  • The option to select terrain vector databases in .tdv format is now available in the list of supported file formats under the menu File > Open Database.
  • Rays can now be displayed for time-variant projects with a stationary simulation.
  • New EMC specification files can now be created in ProMan.
  • The input and output frame rates of a ProMan animation are now set to be equal. Previously, only the output frame rate was specified, and a constant value was used for the input frame rate. A mismatch in input/output frame rates could have resulted in some snapshots of the animation not being captured.
  • Computation filter settings, specified under global settings, are now applied also to mobile-station (RunMS) results.
  • Fixed a bug that resulted in the propagation path not being correctly displayed in 3D view for paths longer than 20 km.
  • Fixed a bug that resulted in a misalignment between the computed results and the defined prediction area, for a rural scenario project with the topography defined in geodesic coordinates.
  • Fixed a bug in the determination of line-of-sight pixels for projects consisting of a directive transmitter and solved with the dominant path model or as ray tracing models (SRT/IRT) when the contribution of the direct ray is disabled.
  • Added support for the use of database files with multiple extensions when creating a new ProMan project.
  • The interference between antenna components is now correctly considered only when the antennas are fed by different transmitters at the same carrier frequency. Previously, antenna components were treated as though they were fed by different transmitters, leading to incorrect interference results (low SNIR values).
  • Translation and rotation of a receiver, along a trajectory, are now considered for power azimuth spectrum computations.
  • The effects of the curvature of the earth is now more accurately considered in the deterministic and empirical two ray models in rural scenarios.
  • Fixed a bug that not all transmitters were listed in the network planning ASCII result files.
  • Improved ProMan's user-friendliness in case a user tries, by mistake, in a network planning project without a pre-defined air-interface file, to assign a carrier frequency to a transmitter before any carriers are defined. It used to be difficult to exit the particular dialog in that situation. Now the user is informed about what is missing and on which dialog the carriers can be defined.
  • Fixed an issue that in a network planning project with components, depending on the air interface, some results could not be displayed.
  • Fixed a display bug in the 3D view that resulted in an offset in the elevation profile of the displayed results of a tunnel being introduced when disabling topography from the display settings.
  • While rural ray tracing is supposed to include interactions with the terrain only in the vertical plane, it used to find and use wedges in the terrain everywhere. This incorrect behaviour, which could increase the simulation time dramatically, is corrected. For 3D interactions with terrain, other simulation methods remain available.
  • Fixed a bug where the ProMan project icon (top left, next to the File menu) had incorrect options and could sometimes make ProMan crash.

WallMan

Feature

  • Added support for digital terrain elevation data in the .dt3 format.

Resolved Issue

  • Saving building shapes for an imported indoor database can take significant time while it is only necessary when preparing a hybrid urban/indoor scenario. Therefore, it is now disabled by default. This option can be activated on the File > Save Database As dialog.

AMan

Resolved Issue

  • Fixed a bug that resulted in a crash when combining antenna patterns with the multiple antenna configuration feature in AMan.

Application Programming Interface

Features

  • Added support for in-model parallelisation in the WinProp API for propagation predictions, network planning as well as database preprocessing.
  • Added examples demonstrating multi-threading with the WinProp API.
  • Added an example demonstrating urban database preprocessing using the WinProp API.

Resolved Issues

  • Updated the path of the input directory in the distributed WinProp API usage examples.
  • The database conversion function, WinProp_Convert, now converts indoor databases to the .idb binary format by default.
  • Error codes returned by the functions of the WinProp API now match the description given in the reference documentation.
  • Specific error messages are now returned from the WinProp API in Linux in cases where a generic Unknown error was issued before.