This chapter describes configuration and use of the AnalystTM 3D Electromagnetic simulator. The Analyst simulator is compared to the AXIEM simulator, and results and post-processing are also discussed. The Analyst simulator uses the frequency-domain Finite-Element Method (FEM) on a tetrahedral mesh in a three-dimensional volume that may include metal, air, and dielectric. It adaptively refines the mesh over several iterations to optimize the quality of the solution for a given computational burden. The Analyst simulator makes no assumptions or approximations regarding the geometry, so you can use it for essentially any model. You can generally use it on structures set up for AXIEM simulation with no (or minimal) model changes, although it has a longer run time for most problems and uses more computer memory and resources. It also produces larger datasets than AXIEM, particularly when field output is turned on.
This chapter focuses on using the Analyst 3D Editor from the Layout Editor within the NI AWR Design Environment suite. Documentation for the Analyst 3D Editor is found within the 3D Editor environment. In the 3D Editor, you can access Help in several ways:
Press the F1 key
Click the question mark symbol located at the top right of the 3D Editor window
Click thebutton located at the top left of the 3D Editor window, then choose
This section is for experienced users who want a reminder of the items to consider when setting up an Analyst simulation.
Are the ports properly configured? See “Analyst Ports” for details. A common mistake is to include internal implicitly grounded ports. These are not supported and cause the simulation to fail
Are the proper boundary conditions set and is the boundary spaced far enough away from your geometry? See “Specifying Simulation Boundaries” for details. Both of these are common mistakes that can have dramatic effects on your answers.
Have you set a good AMR frequency for your structure? See “Adaptive Mesh Refinement Options” for details. This is especially important for filters or resonant structures.
Have you set the appropriate frequencies to save the field outputs? See “Solver Options” for details.
Have you changed your port impedances for the EM document? See “Understanding Port Impedance” for details. A common mistake is to change these values and expect them to change the simulation results.
You can follow these guidelines when determining whether to use the Analyst 3D or AXIEM simulator.
Use the Analyst 3D simulator for:
All true 3D problems, for example, waveguides and connectors.
3D pCells, for example, bond wires and ball grids.
“Planar” problems with 3D effects such as thick metal problems where skin depth is comparable to thickness (for example, inductors), many ground planes (3D FEM may have speed advantage), metal enclosures, and dielectric bricks.
Use the AXIEM simulator for:
Problems that contain only planar metal and dielectrics and open boundaries, particularly when modeling the metal as zero-thickness. The presence of thick metal and vias slows AXIEM, but it is often faster than 3D FEM for such problems.
Planar radiators; these are more efficiently analyzed with AXIEM, as its formulation assumes radiation boundaries, whereas in 3D FEM, radiation boundaries must be synthesized using special boundary conditions or computationally expensive materials.
You need to consider the model configuration, basic setup parameters, meshing, and basis set differences when setting up a problem for AXIEM or Analyst simulation.
You must specify overall boundary for 3D calculations (since the air volume around the structure is explicitly meshed and included in the finite-element simulation).
Analyst simulation supports a subset of the port types AXIEM supports, but also allows the use of wave ports, which AXIEM does not support. A wave port is a planar surface, usually oriented so that signal conductors or waveguide crosses the surface at right angles, and is analyzed for propagating modes that are subsequently used to excite the port in the full EM solve.
A variety of 3D pCells exist that only the Analyst simulator can use.
The following are current Analyst limitations:
No DC solution: “Low” frequency is used in place of true DC. The default value for Minimum Solved Frequency is 10 MHz. The solution at (and above) the Minimum Solved Frequency is extrapolated to any frequency below this value. As a result, losses and other quantities may not be correct at DC with this approximation.
No lumped port de-embedding.
No support for arbitrary orientation of child EM structures.
To convert an AXIEM project to an Analyst 3D project:
Right-click the project name in the Project Browser and choose Set Simulator. In the Select a Simulator dialog box, choose AWR Analyst 3D EM - Async.
By default, a boundary shape draws slightly larger than the geometry of the structure. Resize the boundary appropriately for your structure. For more information on choosing the boundary size, see “Boundary Size”. You typically define the boundary size so it is 10x the thickness of your dielectric, away from any conductors on the sides without wave ports. If you use wave ports, you must have the boundary touch the ports on those sides.
Ensure that the port types are correct. Analyst simulation supports the same ports as AXIEM except for implicitly grounded internal ports. See “Analyst Ports” for details.
Right-click the project name in the Project Browser and choose Options. In the Options dialog box on the Mesh and Analyst tabs, set the desired options. Typically, the defaults are sufficient.
The geometry that Analyst simulation uses is defined with the same editor that defines geometry in any other EM solver (such as AXIEM or EMSight) in the NI AWR Design Environment suite. For example, any structure defined for AXIEM can easily be used as an Analyst document. Analyst documents also allow support for additional 3D constructs. There are 3D parameterized cells (pCells) you can use to add 3D components such as bond wires, ball grid arrays, tapered vias, and others. Within an Analyst EM document, you can define extruded 2D shapes and vias as conductors or dielectrics (in AXIEM, only conductors are supported).
You can control the drawing grid for Analyst EM documents. Double-click the Enclosure node under an Analyst document to display the Element Options - ENCLOSURE Properties dialog box , then click on the Enclosure tab. The Grid_X and Grid_Y parameters set the drawing grid for the document.
When drawing shapes, you use the mechanisms used for any of the other simulators. See “Drawing Shapes in EM Structures” for details on drawing shapes in the EM Layout Editor.
Since Analyst is a full 3D FEM solver, it can simulate finite dielectrics. In the STACKUP you define dielectric layers that are continuous over the entire structure. You do not need to draw shapes to define these layers. If you want to define a smaller region that has different dielectric properties, the process is the same as adding a conductor shape except that the material definition is a dielectric material instead of a conducting material.
Frequency-dependent materials are listed in “Frequency-dependent Material Definitions ”. For a given material, the individual properties may be frequency-dependent or constant, and they can also be either scalar or anisotropic, in any combination.
For a true 3D FEM solver like Analyst, all space within the simulation boundary must be completely defined by a collection of non-overlapping 3D solids. The detail of creating the non-overlapping shapes is handled automatically using a few simple rules that define which material takes precedence when two 3D shapes overlap completely or partially. You do not typically need to be concerned with this automated process, however, in certain situations where extruded dielectrics are used, or where conductors with different material properties overlap, it is helpful to understand how the rules work. The rules are based on a spatial priority, where bodies with higher priority take precedence over bodies with lower priority:
Extruded dielectric layers defined by the stackup are considered background layers; these have the lowest priority.
Extruded dielectric shapes and vias have the next highest priority. Dielectrics with higher Er have a higher priority than lower Er dielectrics.
Conductors and vias have the highest priority. Conductors with higher sigma have a higher priority than lower sigma conductors.
The following scenarios illustrate the priority rules:
When normal conductors are created from extruded 2D shapes within the EM document, these conductors automatically 'cut' away the space from the dielectric layers defined by the stackup because the conductors have a higher priority than the background dielectrics.
If you add an extruded dielectric shape, it also 'cuts' away the space from the dielectric layers because it also has a higher priority than the background dielectrics.
If a conductor overlaps the same volume as an extruded dielectric, the conductor 'cuts' out the space from the extruded dielectric.
If a conductor with higher sigma overlaps a conductor with lower sigma, the higher sigma conductor 'cuts' out the space from the lower sigma conductor.
For more information, see “Material Priority in Hierarchy”.
When the geometry contains overlapped materials and structures, or when you are not sure how the solver will interpret the geometry, it is helpful to see how the intersecting materials are treated before the system is simulated. To view the structure as it is input into the simulator, choose View > Visual > Merged 3D Model. The 3D geometry displays with all pre-processing steps completed. You can use this view to verify that the structure under test is designed as intended.
3D EM elements are specialized elements for use with the Analyst 3D Electromagnetic simulator. For these elements, layout parameters control the 2D projection of the element in the Layout Editor (such as x-y location and orientation) while element parameters control the 3D properties of the element.
To add 3D elements to an Analyst EM document layout:
In the Elements Browser, expand the 3D EM Elements node as shown in the following figure.
Select the 3D element you want to add, then drag it from the Element Browser to the EM layout window. NOTE: You must be online to access these elements from the NI AWR website. See “Offline 3D EM Elements” for details on how to use these parts when offline.
After dragging to the layout, you are in placement mode. A ghost image of the layout displays.
Click to place the element, then right-click to rotate it 90-degrees. To flip the cell about the y-axis, hold down the Ctrl key and right-click. To flip the cell about the x-axis, hold down the Shift key and right-click.
Each element knows on which EM layer it is located. When you add an element, it is added to the active EM layer. Click the Layout tab to open the Layout Manager and set the active layer by clicking the arrows on the right of the EM Layers pane to open it. As shown in the following figure, the EM Layer sets the active layer.
You cannot add 3D elements to the top of the structure, so if this layer is set to "1" or "EM layer undefined", the pCell is added to the bottom layer. The following sections describe how to change a pCell's EM layer after you add it.
After adding the element to the 2D layout, you can edit the 2D properties of the pCell the same way you edit any other pCell layout. You must then configure the 3D properties of the element and the Z position of the cell in the stackup.
Each 3D element has parameters that control its basic 3D properties. For example, a bond wire has parameters for loop characteristics, and a capacitor has the dielectric thickness. You can edit these properties by selecting the element in the EM layout, right-clicking and choosing Element Properties. The values of all of the parameters relating to height are relative to the Z position specified for the element.
In this dialog box you also specify the material properties for the element such as dielectric constants and metal conductivity. The NI AWR Design Environment software then creates all the appropriate materials needed to simulate these elements. This is different than other shapes where you must set up the EM mapping to map a layer name to an EM layer and material.
You must also specify the Z position for each 3D element. Select the element and click theor buttons on the toolbar to move the pCell up or down one EM layer.
You can also select the element, right-click and choose Layout tab, under Z Position specify on which EM layer to place the element, or enter a Z Offset. Typically you use the EM layer for the move and only use the Z offset if you need more control over the pCell movement. You can also rotate the pCell about the x, y, or z axis.to display the Cell Options dialog box. On the
In an NI AWR Design Environment layout, 3D elements do not have to be set up the same as other pCells. These elements only exist in EM documents, and all the material information for conductors and dielectrics are contained in the elements. Layers and materials are automatically created for these elements. The layers display in the Layout Manager on the Drawing Layers pane. These layers control visibility in the 2D and 3D shapes for these elements. The Visibility by Material/Boundary pane shows the auto-generated dielectric and/or conductor materials used for simulation. The following are the drawing layers generated by a subset of the 3D elements:
Bond Wire: BondWires
Tapered Via: TVIA
Thin Film Capacitor: CHPCAMP for the cap metal and CHPCAPD for the dielectric.
Each of the 3D pCells defines its conductivity relative to gold. If RHO is 1, the material name is "Gold". For any other Rho, the material name is "Gold_x_A" where A is the number set to Rho. For example, if RHO is set to 0.7 the material is "Gold_x_0.7". The thin film capacitor also creates a dielectric material with the syntax "Die_ER_A_Tand_B", where A is the relative dielectric constant (Er) and B is the loss tangent (Tand) set as the element parameter on the pCell.
NI AWR creates and maintains many generic 3D parts for customer use. These parts are primarily accessed via the NI AWR website. NI AWR may need to update 3D parts as issues are discovered and fixed. This section describes how to determine if an update is available, and how to update the part.
A script that checks for updates is available by choosing.
When you run the script, a new data file is loaded into the project and opened for viewing. It lists any parts that need an update, with information about what has changed and what may be needed to update that part. The following figure shows a window displaying an available update.
The file name includes the date and time. The Browser Path line displays the path needed for the update process. If all parts are current, the window displays as follows:
To update the 3D parts:
Open an EM document that contains the 3D part.
View the schematic of the part by clicking thetoolbar button.
NOTE: If this toolbar is not visible, with the 2D EM layout active, right-click in the current toolbar and choose EM Design.
Select one of the subcircuit instances in the EM schematic that has the NET parameter of the part being updated.
In the Element Browser, browse to the location of the part as listed in the update window. For example, the previous example shows: Browser Path = BP:\3D EM Elements\Libraries\3D Parts\Test Versioning\Test. Right-click the element and choose.
Download and configuration of the new part might take some time.
Note the following when updating a 3D element:
The update mechanism does NOT work if you rename the EM subcircuit for the part. If renamed, a message box similar to the following displays.
The update mechanism updates ALL instances of the same subcircuit, not just those in the current EM document.
The update mechanism should maintain all the other properties of the part, such as the parameter values and the shape properties (for example, EM layer and rotation).
On a computer with internet access, you can directly download an offline library of 3D EM elements. On a computer without internet access, you can install a local copy of the library.
To download the 3D EM Elements library from the NI AWR website, choosewhile online. A copy of the 3D EM Elements library is saved locally for your use when you are offline.
If you are offline and try to use a 3D EM element from the NI AWR website library, the following window displays.
Optionally select a check box for one of the options and then click “Updating 3D Parts ” for details. You can choose to update the cached copies.. Note that your cached copy may not be the latest version. You should periodically run the "Check_For_Update_3D_Parts" script when you are online to ensure the latest version. See
A local copy of the 3D EM Elements library is available from the Downloads page of the NI AWR website. Once downloaded, you can install this library on computers that have no internet access. Note that the "Check_For_Update_3D_Parts" script does not work for local library parts; it only works for downloaded or cached parts.
To install a local copy of the 3D EM Elements library:
On a computer with internet access, download the "AWR 3D Parts
Library" from the NI AWR website under the Vendor Libraries section of the
Downloads page. The download is a
that you can transfer to other computers.
To install the library, extract the contents of the .zip file
3D EM Elements directory. The path
to this directory is operating system dependent. To find this
directory, choose Help > Show
XmlUser in the list of directories. The
3D EM Elements directory is one of the
sub-directories listed in the new window that displays. You
should extract the
.zip file so the
AWR_local_3dparts_xx.xml file is
located directly in the
3D EM Elements
Start a new NI AWR Design Environment session to use the newly installed local library. The library displays under the 3D EM Elements > Libraries node in the Elements Browser.
You can use the AWR 3D Editor to make your own 3D parts. To make your own 3D part:
Right-click the EM Structures node in the Project Browser and choose .
Enter the structure name and select an Initialization Option for the structure, then click the button.
Right-click the new EM structure and choose.
The 3D Editor opens and you can create your arbitrary 3D structure. The details for using the editor are provided in the editor documentation. See “Analyst 3D Editor Help” for details on accessing Help in the 3D Editor.
You can build a library of the 3D parts created in the NI AWR Design Environment 3D Editor. This process requires only simple file management and some XML file editing.
This section describes how to create a library of 3D structures using XML. When building your library, you need to create a folder structure. You will have XML files to pull the library together and additional files that define your 3D geometry. The following example shows the process to create an 3D part with a project that contains an arbitrary 3D EM Structure modelling a SMA connector.
Create a folder named "3D XML" on your PC.
Inside the "3D XML" folder create a subfolder named "connectors".
Export the arbitrary 3D EM structure by right-clicking the arbitrary EM document under the EM Structures node and choosing .
Choose the "connectors" folder, specify "SMA" as the File name, and then click .
Note that you can only save the file as a
file, however, this creates five files with
as shown in the following figure. You need all of these files to create
Open any text editor (such as Windows® Notepad), type the
following content, and save the file as
connectors.xml in the "3D XML" folder. For more
information on NI AWR’s XML Schema description, see Appendix A, Component Libraries.
<?xml version="1.0"?> <XML_COMPONENT_DATA xmlns="urn:awr-lib-data"> <COPYRIGHT>AWR</COPYRIGHT> <SUMMARY>AWR 3D Parameterized Models</SUMMARY> <FOLDER Name="Preferred Connectors" Icon="Connectors"> <COMPONENT Name="SMA Edge Mount"> <MODEL/> <DESC>Female Jack Connector</DESC> <PARTNUMBER>SKDSMA2P</PARTNUMBER> <SYMBOL>SMA@EM3D.syf</SYMBOL> <HELP>HelpSMA.pdf</HELP> <DATA DataType="awrschematic" Inline="no" LinkToFile="no">connectors/SMA.gml <PARAM ParamUnits="length" Name="ContactLen">508e-6</PARAM> </DATA> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >connectors/SMA.apz</SUBFILE> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >connectors/SMA.sat</SUBFILE> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >connectors/SMA.prop</SUBFILE> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >connectors/SMA.GP.m3d</SUBFILE> </COMPONENT> </FOLDER> </XML_COMPONENT_DATA>
When this XML library is added to the NI AWR Design Environment suite as described in the following steps, it displays as shown in the following figure:
In the XML, the <SUMMARY> and <FOLDER> sections are optional. The library name, as shown in the previous example, My Custom Connectors is defined by the XML file name. The subnode Preferred Connectors is defined in the <FOLDER> section.
The <COMPONENT> section is mandatory and is used to define the type of component. For example, SMA Edge Mount, which displays under the Models column.
The <MODEL> section is ignored since it is not used in this library.
The <DESC> section is mandatory and the value describes this component. For example, "Female Jack Connector" which displays under the Description column.
The <PARTNUMBER> section is optional and is used to define the part number for this component. If defined, it displays in the Element Options dialog box for the component.
The <SYMBOL> section is mandatory. You can use existing symbols
from the NI AWR Design Environment suite, for example,
The symbol is displayed next to the component name in the Element pane
as shown in the following figure, and is placed in the EM schematic
window when the component is added to the EM document.
The <HELP> section is optional. It points to a file that contains Help for this component. The file is specified as a relative pathname or full URL.
The <DATA> section is mandatory. NOTE: For this type of library the DataType must be set
to "awrschematic" and the value must be the
file. Set Inline="no" and LinkToFile="no".
The <PARAM> section is optional, and defines a parameter for the component. There is a PARAM entry for each of the component's parameters. Name is set to the parameter name, such as "ContactLen" or "CoaxLen". ParamUnits is set to, for example, length, and the value must be in base units (for length, it is meters). Other ParamUnits supported are angle, resistance, capacitance, conductance, voltage, current, and power.
The <SUBFILE> section is mandatory. There must be a separate
<SUBFILE> section for each of the
.GP.m3d files. Set DataType="downloadonly",
Inline="no", and LinkToFile="no" for all of them.
connectors.xml file is ready, add the
file to your
XmlUser\3D EM Elements folder. In this
example, the file paths define in the <SUBFILE> section of the XML
file were defined relative to the XML file location, so you also need to
copy the "connectors" subfolder to the
To find the
XmlUser folder, choose Help
> Show Files/Directories to display the Directories
dialog box. Double-click the
XmlUser folder, as
shown in the following figure. Open the
Elements folder and add the XML to this folder.
Start the NI AWR Design Environment program or open a new project to update the Element Browser.
In the Elements Browser, expand the 3D EM Elements > Elements node. The newly added library displays.
Click the library name to display its elements in the bottom pane of the Elements Browser, as shown in the following figure.
If you are building your own library of parts, you can add versioning to the 3D part use the “Updating 3D Parts ” for details.script that checks for updates. See
Version information is stored as text on the EM schematic for the Arbitrary 3D EM document added to the NI AWR Design Environment suite. This 3D part information is accessible by thescript. You must add the versioning information before exporting arbitrary 3D EM structure when creating an XML library. To add versioning information:
With the Arbitrary 3D EM document window active (your 3D part), view the schematic of the part by clicking thebutton on the toolbar.
NOTE:: If this button is not visible, right-click the toolbar and ensure that the “3DEM Layout” toolbar is selected for display.
To add version information, choose
Version:1". You should
increase this version number when there is an update.
Export the arbitrary 3D EM structure by right-clicking the arbitrary EM document under the EM Structures node and choosing. Proceed with the XML library creation process.
To update the version on an existing part, repeat the same procedure,
except increment the version number in the EM schematic text. Replace the
.GP.m3d files with the newly exported
Thescript determines the version of a 3D part stored in the project from the versioning text in the 3D part EM schematic. The script also determines the XML file location from which the 3D part came. The script then queries this XML file looking for the current version and any release notes, and then determines whether the part saved in the project is up to date.
NOTE: If you change the structure of the XML, the update mechanism does not work unless you write scripting code to map the old locations to the new.
In order for thescript to work, you need to add the versioning information to the XML file defining the library part. You must update the version number set in the XML file when the version number in the 3D part EM schematic is updated. When running the script, the "Version" value set in the XML is checked against the "Version:X" number of the 3D part used in the project to determine if an update is available.
The following example shows a part configured with version number and release information, where relevant information for the update is listed in the last four <PROPERTY> lines.
<COMPONENT Name="Test"> <!-- Model name appearing in project list --> <MODEL>test</MODEL> <DESC>Test Part</DESC> <PARTNUMBER/> <SYMBOL>ParRLC@system.syf</SYMBOL> <HELP/> <DATA DataType="awrschematic" Inline="no" LinkToFile="no" >resonators/DISK_RESON.gml</DATA> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >resonators/DISK_RESON.apz</SUBFILE> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >resonators/DISK_RESON.sat</SUBFILE> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >resonators/DISK_RESON.prop</SUBFILE> <SUBFILE DataType="downloadonly" Inline="no" LinkToFile="no" >resonators/DISK_RESON.GP.m3d</SUBFILE> <PROPERTY OnInstance = "no" Name="Version" Value="3"/> <PROPERTY OnInstance = "no" Name="RN1" Value="Updated Scaling for the H Parameter."/> <PROPERTY OnInstance = "no" Name="RN2" Value="Added proper help link."/> <PROPERTY OnInstance = "no" Name="RN3" Value="testing update."/> </COMPONENT>
The "Version" <PROPERTY> line is required for the update script, where "Version" case is not important. You can then optionally add release note <PROPERTY> lines using the syntax "RNX" where X is a number indicating the changes from X-1 to X. For example, "RN1" is the change from version 0 to version 1; case is not important.
NOTE: The OnInstance attribute is user-defined and specifies whether or not the information is stored on the instance in the NI AWR Design Environment program. In this example it is not, because the information is not needed on the part and it increases the size of the NI AWR project file.
If you change the Name to use User:Version and omit the OnInstance attribute, this information displays on the Element Options dialog box User Attributes tab. You can use "User:*Version" syntax and the User Attribute is read only. See “Adding User Attributes in XML Files” for details.
You can directly import GDSII and DXF layout formats to an EM structure. See “Importing GDSII/DXF as an EM Structure” for details.
You can design EM structures hierarchically using other EM documents as submodels. This approach greatly simplifies the setup for complex dielectric stackups. If you have a structure on a GaAs chip that has two dielectric layers, GaAs and Air, and you want to use this chip on top of a board or package with an interconnect, you typically specify the PCB dielectric layers (two layers in the simple case, one for the board and one for Air). You can then use the chip EM structure as a subcircuit in the PCB structure with no additional dielectric layer setup. The simulation for the PCB resolves all of the dielectric layers and conductors for any shapes in the PCB structure as well as the chip subcircuit. This allows the individual pieces to be simulated separately and then quickly integrated to simulate together with minimum setup.
You can add any EM structure as a subcircuit to any other EM subcircuit. With an EM structure 2D layout window active, choose “Working with Hierarchy” for more information about EM hierarchy.or click the button on the toolbar. Select the subcircuit you want to add, then place it on one of the EM layers in the current structure. The rules for the initial placement and editing of the Z position of the subcircuit are the same as those for 3D pCells. See
Each Analyst structure must have specified boundary conditions. Only the shapes inside the boundary are simulated, and all ports must be inside or on the perimeter of the boundary. The port type is determined in part by the position of the port with respect to the system boundary; see “Analyst Ports” for more information. The boundaries are defined by the shape in the 2D layout of the EM structure that encompasses the rest of the geometry. You choose the boundary condition to use for the boundary top, bottom, and sides. Boundaries without assigned conditions default to the Approx Open boundary condition.
When you first create an Analyst EM structure or convert an EM structure to Analyst, a boundary shape is added by default. There are two options for editing boundary shapes:
Edit an existing boundary shape by double-clicking it to place it in edit mode, then click and drag any vertex or edge midpoint.
Draw any shape or select any shape in the EM layout. Choose Create Simulation Boundary button on the toolbar to make the shape a boundary shape.or click the
When using this command, more than one boundary shape might be defined. If so, an error message displays and no simulation occurs. You should delete any extra boundary shapes prior to simulating.
NOTE: If you have more than one boundary shape, you cannot perform Boolean operations on them to make a more complex boundary shape. (You might have already set different boundary conditions on each edge of the boundary shape, and the result of the Boolean operation would not know how to keep those changes.) Therefore, if you want to make a complex shape for a boundary condition, first make the shape on any other layer. In this layer, you can perform any complex editing such as Boolean operations or notches, and then use the first option listed above to convert the shape to the boundary condition.
The boundary size can significantly impact the quality of the simulation results, particularly if the boundary is too close to metal shapes. For non-radiating structures, if the boundary is far enough away that the fringing fields are near zero, the boundary is electromagnetically isolated from the structure of interest and the solution should be insensitive to the choice of boundary condition. Generally, for simple cases you can place the sidewalls and top surface at least 10x the substrate thickness away from the nearest metal. At this distance the solution is largely unchanged whether you use PEC, PMC, or approximate open boundary conditions. For more complex geometries, you may need to experiment to determine minimum safe boundary sizes. This spacing requirement applies to all sides of the domain that do not contain ports. On sides with ports, the field is forced to a particular mode solution and the choice of boundary condition has no impact on the simulation results.
For radiating structures, the PML boundary condition gives the most accurate results, but the less-accurate Approximate Open boundary condition may also be used in the design phase. Place PML boundary conditions on either the boundaries of a rectangular volume or on the exterior of a spherical volume. Approximate Open boundary conditions may be placed on boundaries of any shape. Generally the nearest boundary walls should be, at minimum, one quarter of a wavelength away from the nearest metal, but not more than a few wavelengths away.
Boundary conditions are chosen based on a given problem. Fields are modeled only within the enclosure, and treatment of the fields on the boundary of the enclosure is determined by the choice of boundary conditions. In general, boundary conditions can specify values of particular components of the electric or magnetic fields, or they can define relationships between E and H.
The following table summarizes the types of boundary conditions that the Analyst simulator supports. The table also indicates the effect that each boundary condition has on both the electric and magnetic fields at the boundary. Further details on each type of boundary condition are listed after the table.
|Boundary Condition Type||Effect on
||Boundary Size Recommendations||Use Notes|
|Perfect Electric Conductor (PEC)||
||Typically used to model a structure in a conducting enclosure, so define the size to match the size of the actual device.|
|Perfect Magnetic Conductor (PMC)||
||5-10x substrate height away from metal shapes.||An alternative to Approximate Open for non-radiating structures.|
||Typically used to model a structure in a conducting enclosure, so define the size to match the size of the actual device.||These choices have names that correspond to material definitions such as Tantalum or Cap Bottom.|
||5-10x substrate height away from metal shapes.||Used on non-radiating structures to approximate free space, and on radiating structures when speed is paramount.|
|Perfectly Matched Layer (PML)||see below||Place > λ/4 away from metal shapes.||Intended for use in true radiating structures. Computationally expensive.|
||n/a||Used only when geometric and electromagnetic symmetry are present. Must bisect all ports.|
||n/a||Used only when geometric and electromagnetic symmetry are present. Must bisect all ports.|
The Perfect Electric Conductor (PEC) boundary condition defines the material with infinite electrical conductivity, so the tangential component of the electric field and the normal component of the magnetic field go to zero on the boundary. In many cases this idealized material is a good approximation for highly-conductive metals, and allows for faster computation time.
The Perfect Magnetic Conductor (PMC) boundary condition is assumed to have infinite magnetic conductivity, so the tangential component of the magnetic field and the normal component of the electric field go to zero on the surface.
If you choose to terminate your geometry with a material of finite conductivity, you have applied the Impedance boundary condition. The Analyst solver uses both the conductivity and the thickness of this material to determine the impedance at the boundary.
The Approx Open boundary condition is a zero-conductivity impedance boundary condition with the impedance of free space. It is a good choice for non-radiating systems that are not bounded by conductors. It may also be used as a alternative to PML, when some level of accuracy may be sacrificed for solver speed. As noted in the previous section, you should ensure that the boundaries of such systems are far enough away from the structures of interest that the fringing fields are attenuated to near-zero at the boundary. The approximate open boundary condition is useful because it does not fully reflect the low fields at the boundary, as PEC and PMC do, although some reflections still occur.
The Perfectly Matched Layer boundary condition defines an auxiliary mesh that adjoins the main FEM mesh at the surface that has the PML boundary attribute. This auxiliary mesh is filled with a graded lossy medium that is impedance-matched to the adjacent main mesh, so that incident fields penetrate the PML mesh with low reflection, and those fields are attenuated in the PML material. Field attenuation within the PML depends on the incident angle of the fields that intersect the PML boundary. Normally-incident fields experience optimal attenuation, while glancing fields experience less attenuation. Materials that intersect the PML go to infinity. PML may be applied to either planar or spherical surfaces, but not both within the same geometry. Using PML boundary conditions on a spherical boundary may result in a faster simulation than using PML on a rectangular boundary, because the auxiliary mesh will usually add fewer elements to the system than it would in the rectangular case. In the 3D editor, define a spherical simulation domain instead of a rectangular one, and apply the PML to the exterior surface. See “Using Boundaries Defined in Arbitrary 3D EM Structures” for more details.
Electric Symmetry and Magnetic Symmetry boundary conditions reduce the computational size of a problem, or allow you to get more accurate and detailed mode data for a set amount of computer resources. Use of symmetry also thins the mode spectrum, as it prohibits modes that do not exhibit the specified symmetry. Although the electric symmetry plane is electromagnetically equivalent to PMC, and the magnetic symmetry plane is the same as a PEC, the Analyst solver treats them differently from PMC and PEC planes. Symmetry planes are intended to let you decrease the physical size of a system by exploiting natural symmetries in the fields. You cannot specify symmetry on two surfaces that are parallel but offset in their normal directions. Symmetry planes do not have to be orthogonal to each other, although only one pair of planes is allowed to be non-orthogonal. If the combination of planes is too complex for the solver to compute a volume fraction, it outputs a warning and ignores symmetry in field normalization and mode parameter calculations. When using symmetry planes with spherical PML, the center of the PML sphere must be coplanar with the symmetry plane. If you have a symmetry plane in a simulation with either a wave or lumped port, the symmetry plane must pass through the port. So, you can use symmetry planes to increase the size of a port and thus show the symmetry within the port solution, but you cannot use symmetry planes to duplicate ports.
There are two options for editing boundary conditions:
Select the boundary shape, right-click and choose Boundary Conditions tab.to display the Properties dialog box, then click the
Double-click the Enclosure node under the EM document in the Project Browser to display the Properties dialog box, then click on the Boundary Conditions tab.
By default, each side uses the Side Boundary setting. You can set each side individually by clearing the Use Side Boundary check box and then setting each edge of the boundary independently below this option. When doing so, open the 2D EM layout so you can see the edge you select highlighted, as shown in the following figure.
NI AWR recommends visualizing the boundary conditions for an Analyst structure in the 3D view to make sure they are set up as intended. With the 3D view open, click thebutton on the toolbar.
An "EM3D_SURF_BC" annotation is added below the EM document in the Project Browser. You can double-click the annotation to see the available options. See “Surface Boundary Conditions for 3D Models: EM_3D_SURF_BC” for details. Click the toolbar button again to turn off the boundary condition display. The Layout Manager Visibility by Material/Boundary pane can quickly show/hide boundaries by material type, which is useful for verifying that all the material properties are set up as expected.
You can use the boundary shape to simulate a subset of the drawn geometry. Only shapes that are located inside the boundary shape are simulated. Any ports used for simulation need to be inside or on the edge of the boundary shape or a simulation error occurs. You may have to add shapes to attach ports when moving the boundary condition.
The following example demonstrates this concept. This figure shows a simple line where the boundary shape encloses the entire geometry.
You can edit the boundary shape to only include some of the geometry as shown in the following figure. Here, port 2 is outside the boundary area so a simulation error occurs.
Port 2 is added to a valid region in the following figure, so the structure now properly simulates.
If you view the mesh of this structure, you see that only the region inside the boundary shape will mesh.
You can define an arbitrary 3D EM structure entirely in the NI AWR Design Environment 3D Editor, including the boundary conditions for the simulation. If the boundary conditions are defined in the arbitrary 3D EM structure, the extruded 2D enclosure might not be required when using this structure in an Analyst 3D EM structure. If so, you can choose to ignore the enclosure defined in the Analyst document by right-clicking the EM structure in the Project Browser and choosing Options to display the EM Options dialog box. On the Analyst tab, click the Show Secondary button, select the Ignore Enclosure check box, and then click OK.
See “Adding EM Ports” for information on how to add and edit ports in the EM Editor. This section provides additional information on how ports are modeled and used in Analyst simulations. In general, when using the Analyst solver you can assign a port to any conducting edge in the system. All output quantities are defined in terms of port number.
The Analyst solver supports both wave and lumped ports. You must explicitly specify the port type. Both port types can be placed anywhere in the geometry. The extents of the wave port plane can either be determined by the geometry or you can specify them directly by choosing the Wave Custom Size type. The lumped port type determines the presence or absence of an explicit ground reference.
Excited wave ports introduce energy into the simulated fields, in the form of the lowest-order eigenmode of the structure defined on the port plane. For each simulation frequency, the solver performs an eigenanalysis of the system on the port plane, and each of the resulting modes determines the amplitude profile of the excitation for that frequency. In the "Full" solve, the finite-element matrix equation is solved once for each port eigenmode, so that the field on the port plane is enforced for that eigenmode. The technique used in wave ports does not introduce parasitic components to the fields, as may happen with lumped ports, so there is no need to de-embed wave ports.
Wave ports must be specified on external boundaries of the domain, except in the special case described in “Internal Wave Ports”. If the port type is Wave, the port plane extends over the entire side of the enclosure boundary. If the port type is Wave Custom Size, the port plane extends over only the custom extent region. To avoid undesired resonances with the boundary or interactions between field and boundary, you should be sure the port plane boundaries are sufficiently far from the structure to which you assign the port. You may have any number of wave ports defined on a single plane; the Analyst solver computes results for each one independently, accounting for the coupling between them. You may also specify differential port pairs with wave ports, by defining one port with a given positive index, and the other port with the negative of that index. For a given index you may have any number of negative terminals, but only one positive terminal. You may define any number of differential port pairs, as long as each pair is indicated by a unique index.
The Analyst eigenmode solver is optimized for lower-order modes, and it therefore gives best performance on port planes with only a few conductors. Using the Wave Custom Size port can greatly speed up your simulation time without sacrificing accuracy. You can use custom wave port extents to exclude extra conductors from the port solve eigenanalysis. You can also use custom extents to separate wave ports that you know do not interact on the port plane. You should use care with these techniques, to make sure you do not exclude any conductors that are relevant to the desired mode. You also need to be sure the custom region is large enough that the boundaries do not interfere with the desired mode. The boundary conditions at the edges of the custom extent regions are taken from the boundary conditions on the wave port plane.
In the following geometry, custom extents are used to exclude the conductor on the far left from the mode solution. The two conductors on either side of the port are included in the mode calculation.
If you have several wave ports on the same plane and you know they do not interact in that plane, you may use custom extents on them to divide a complex port definition into several simpler port definitions.
If the custom extent of several wave ports touch or overlap, the resulting wave port plane extends over all connected sets of custom extents, and the port solve includes all involved wave ports. This is exactly what would happen if you had not used custom extents, except that the wave port plane is limited to the combined custom-extent region. If you mix Wave Custom Size ports and Wave ports on the same plane, the two types of ports are treated independently in the port solve.
A lumped port introduces excitation by defining a voltage difference between two points for a given port impedance. A lumped port must have a reference point for the port voltage, and thus the port definition must have a corresponding negative terminal or explicit ground reference. In a differential port, the positive and negative terminals are the points that are chosen for the voltage difference. In all other lumped ports, one terminal is on the port, and the other is on the grounding material, as determined by the explicit ground reference. As with wave ports, for a given port index you may define any number of negative terminals but only one positive terminal. In addition, you may define any number of differential port pairs, as long as each pair is indicated by a unique index. In every case, all terminals must be defined on good electrical conductors. If this condition is not met, the solver produces an error and aborts the simulation.
Note that the Analyst solver does not support de-embedding for lumped ports. Nevertheless, to compare results between AXIEM and Analyst simulations, you should de-embed the ports in AXIEM. Due to differences inherent to the simulation methods, the parasitic components introduced by these ports have a much smaller impact on Analyst simulation results than they do in AXIEM results.
Analyst models point ports as very narrow lumped ports that reside at the center of the shape. Lumped port details and restrictions also apply to point ports.
When you double-click a port on an EM document, the Port Attributes dialog box displays and you can edit the port attributes. See “Port Attributes Dialog Box” for more information. You can change the port impedance and power levels. These values are ONLY used when viewing currents and fields directly on the EM structure. They DO NOT affect the simulation S-parameter when plotted on a graph directly from the EM document or when the EM document is used as a subcircuit in a schematic. In this case, the data is always referenced to 50 ohms. If you want to view the results in a different impedance system, you should use the EM structure in a circuit schematic, wire up ports to each node of the EM structure and change the impedances of the ports in the schematic.
The following sections include port usage recommendations for specific structures.
When defining CPW and other structures with multiple ground planes, you must always add the negative terminal port when using lumped ports. The following figure shows the correct CPW lumped port setup.
When using wave ports, only the positive port terminal is required. For GCPW (grounded coplanar waveguide), if the side grounds are connected to the bottom ground with a via, it is important to have the via on the port plane so that the effect of it is included in the 2D cross-section port solve. Otherwise, if there is no via on the port face, the port solve sees four conductors that are disconnected from each other and the mode is different from that of the GCPW.
Internal wave ports can be useful if a port with low parasitics is needed in a geometry that does not easily allow termination of the structure on the geometry boundary. A wave port sources the fields that would exist if the structures on the wave port plane extended to infinity. Because of this, it is nonphysical to use a wave port in the geometry interior, so the internal wave port should be used with care to ensure the geometry is still reasonable.
Internal wave ports are defined differently on waveguides and traces. To define an internal wave port on the end of a dielectric waveguide, simply apply the wave port attribute to the dielectric face at the end of the waveguide as usual. If the waveguide has metal walls, the Analyst solver automatically determines the direction of propagation and extent of the wave port.
To define an internal wave port on metal, you must also define the extents of the port plane. First, apply the wave port to the appropriate interior face or edge inside the enclosure. Then, if you are working with an arbitrary 3D EM structure, define the extents in the 3D EM Editor by applying the same port attribute to a dielectric surface that surrounds the metal edge or face. Define such a surface if one does not already exist. The dielectric surface must coincide with the metal edge or face. This surface, and the port attribute applied to it, tells the Analyst solver how large to make the port plane around the metal structure. If you are not working with an arbitrary 3D EM structure, use the Wave Custom Size port type to indicate the size of the port plane.
In all internal wave ports, the solver adds a PEC backing to ensure the wave is propagated toward the side of the port with the most metal. Therefore, the wave port should be located at the end of the trace or waveguide, and not in the middle.
Using EM subcircuits to define hierarchical EM structures often simplifies designs. See “Creating Hierarchical EM Documents” for more information on this setup. When an EM structure is utilized as an EM subcircuit, it does not automatically simulate when you choose . Instead, an EM subcircuit only simulates if it is the source of a measurement or annotation.
The rules that define precedence when materials overlap, as described in “Material Priority with Overlaps”, also apply to hierarchical EM structures, with the exception of Air layers defined in a subcircuit stackup. Non-air dielectric layers defined in subcircuit stackups are treated as extruded dielectrics when the subcircuit is placed into another EM structure. Air layers, however, are ignored when a subcircuit is placed into an EM structure, and the region previously defined by the Air layer is replaced with material defined at the EM structure level of hierarchy. This special case does not apply to extruded air regions, which are defined by drawn shapes instead of stackup layers.
The following example illustrates how material overlaps are resolved in hierarchical EM structures when there is an Air layer defined in the subcircuit. The following figures show the 2D and 3D views of a child subcircuit defined by two dielectric layers. The top layer is Air, which by default is not set to be visible in the 3D view. The bottom layer is defined with a dielectric constant er = 2.
The following figure shows the resulting structure when the child subcircuit is inserted into a parent structure with a background dielectric constant er = 10. Material overlap rules are applied to the subcircuit as if the subcircuit dielectric layers are extruded shapes, with the exception of the Air layer. Even though the child subcircuit dielectric layer has a lower dielectric constant (er = 2) than the parent (er = 10), the subcircuit dielectric 'cuts' away from the parent background dielectric because extruded shapes have higher priority. The Air layer defined in the child subcircuit, however, is ignored in the parent structure, and the region formally occupied by Air is replaced with the background dielectric of the parent structure.
The following example is similar to the previous example, except that an Air box is drawn as an extrude shape, as follows. The 3D view is identical to the previous figure, because the drawn Air box is created at the same size as the Air layer of the previous example.
The following figure shows the resulting structure when the above child subcircuit is inserted into an identical parent structure as the previous example. Because the Air box is defined as an extrude shape, it is not ignored in the parent structure, and it 'cuts' a region out of the background dielectric.
Open type boundary conditions applied to child subcircuits are not applied when the subcircuit is placed into a parent circuit. Open type boundaries include Approx Open, Perfectly Matched Layer (PML), and Perfect Magnetic Conductor (PMC). With one exception, all other boundaries are preserved when a child subcircuit is placed into a parent structure. For example, you can define a Perfect Electric Conductor (PEC) bottom enclosure and side walls on a child subcircuit. When the child subcircuit is placed into a parent structure, the child ground plane and sidewalls are retained. The one exception occurs when the child subcircuit boundary is contacting Air in the child structure. In that case, when that subcircuit is placed in a parent structure, any portion of the child boundary that was in contact with Air is removed in the parent structure. See “Specifying Simulation Boundaries” for more information on Analyst boundary conditions.
You can set up Analyst EM documents to parameterize the geometry. The NI AWR Design Environment software has several modes of using parameterized EM structures. See “Parameterizing EM Structures” for details.
The NI AWR Design Environment software has a general approach to simplifying geometry before simulating. Often you know details about a specific process where the shapes are needed for manufacturing but can be simplified for simulation. You can write rules to simplify the geometry before simulation. See “Geometry Simplification” for details.
Analyst documents behave the same way as other EM structures in the NI AWR Design Environment software. See “Simulator, Mesh, and Simulation Frequency Options” for details.
You need to set up the mesh, Job Scheduler, and solver options prior to an Analyst simulation.
You can configure various mesh options for each Analyst simulation document. These options only impact the initial mesh generated for use in the AMR (Adaptive Mesh Refinement) process, so the defaults are a good place to start. When different settings are required to, for example, produce a finer initial mesh, some of the options can be set on a shape-by-shape basis.
To access global mesh options, right-click the Analyst document in the Project Browser and choose Mesh tab. To access shape mesh options, select the shape, right-click, and choose to display the Properties dialog box, then click the Mesh tab. The following options are important to consider:to display the Options dialog box, then click the
Model As Zero Thickness - the Analyst solver supports zero thickness shapes where material thickness is specified as 0.0 or when this mesh option is true (either at the document level or on a shape-by-shape basis). To compare AXIEM and Analyst simulation results, ensure that both solvers use the same conductor thickness settings. Note that the default global setting for this option is "Off" for Analyst simulations and "On" for AXIEM simulations.
Size Type - Determines whether the largest allowed element size is defined as a relative or absolute size. Element size is also limited by the AMR frequency and simulation basis set, to ensure sufficient sampling at the highest AMR frequency.
Relative Mesh Size - Maximum element edge length relative to 1/10th the diagonal of the bounding box of the structure. To produce a finer initial mesh, lower this value.
Absolute Mesh Size - Maximum element edge specified in working length units. To produce a finer initial mesh, lower this value.
Size Propagation Factor - Controls element size transition rate from coarse to fine areas. Valid values are between 0.0 and 1.0, with smaller values resulting in faster transitions.
Curvature Refinement Level - Controls the accuracy with which the mesh represents curved surfaces and edges. Valid values are between 0.0 and 1.0, with smaller values more accurately representing curved surfaces and edges. This is important because segmented circular arcs are converted to true circular arcs when the structure is handed off for Analyst simulation.
Anisotropic Curvature Refinement - If selected, curvature refinement is mostly confined to curved directions. This can drastically reduce element counts in structure containing long, small diameter cylinders such as bond wires.
Relative Minimum Curvature Size - Defines a floor for the smallest element to be produced by the Curvature Refinement Level. The value is relative to the global maximum element size.
Enable Small Feature Suppression - Can help alleviate meshing problems particularly with complex and/or imported geometry.
Absolute Small Feature Suppression Tolerance - Tolerance is a length. Recommended values are on the order of 0.0. Making this value larger decreases the chance of meshing failures, however making it too large can alter the geometry significantly.
Enable Local Meshing Controls - If selected, meshing is governed by attributes on the geometry (bodies, faces, and edges) that control the meshing process locally.
The Analyst solver can distribute a single job across multiple processors, with one or more processes allocated to each physical core or processor. Each process is currently single-threaded and accesses only its own block of memory using what is known as a distributed memory model.
To access Job Scheduler options and control use of parallel processing, right-click the Analyst document in the Project Browser and choose Job Scheduler tab. See “Setting Job Scheduler Options” for details.to display the Options dialog box, then click the
You can configure various solver options for each Analyst simulation document. Typically, the defaults are a good place to start, although different settings might be required. To access solver options, right-click the Analyst document in the Project Browser and choose Analyst tab. The following options are important to consider.to display the Options dialog box, then click the
Field Output Frequency - Specifies for which frequencies to save the field results from the analysis. If you do not edit this initially, you need to resimulate to use field annotations.
Solve Type - Choose Full to solve on the entire model, or Ports Only to solve on the ports alone. The Ports Only option allows you to quickly characterize the ports without running the more time-consuming volumetric solve.
Sweep Type - Choose Automatic, Discrete, or GAWE. GAWE is a fast frequency sweep. "Ports only" solves always use the discrete frequency sweep type, so the Sweep Type does not impact those calculations. For details on the different types of frequency sweeps available when using the Analyst solver, see “Frequency Sweeps”.
Basis Set - Controls the level of interpolation for the fields that is used on an element in the mesh. For details see “Basis Sets”.
Minimum Solved Frequency - The lowest frequency solved with the finite-element method employed by the Analyst solver for high-frequency electromagnetic simulations. Below this frequency, simulator behavior is determined by the Extrapolate Toward DC setting. For details, see “Frequency Sweeps”
Solve Inside Conductors - By default, the simulator uses the equivalent impedance boundary conditions on all good electrical conductors to make the simulation faster and less resource intensive. When modeling the skin depth to capture losses is important, change this to Always. Note that this setting results in increased resource usage and often very rapid mesh growth in order to capture skin depth effects.
Extrapolate Toward DC - This flag controls solver behavior at frequencies below the value in Minimum Solved Frequency. If you choose Cubic Polynomial Fit, results at frequencies below the minimum solved frequency are extrapolated from the results above the minimum solved frequency, using a cubic polynomial fit. If you choose None, results at frequencies below the minimum solved frequency are found by solving at the minimum solved frequency. For details, see “Frequency Sweeps”.
Characteristic Impedance Method - This flag specifies how the Analyst solver should calculate the characteristic impedance of wave ports. The characteristic impedance of a circuit port is given by the user-defined port impedance. A wave port's characteristic impedance can be computed using several methods: the Power Current method (computed with P/I2), the Power Voltage method (computed with P*V2), the Voltage Current method (computed with V/I), and for homogeneous waveguides, the Wave impedance. For most single-moded transmission lines the first three methods give similar values, while in waveguides all four options can be quite different and the proper one to use depends on the problem. The Wave impedance calculation is not valid for inhomogeneous solutions. If you select Wave for a simulation that includes either wave ports on traces or inhomogeneous waveguides, the Analyst solver reverts to using the Power Current method for all wave ports.
Remote Simulation for Refine Solution - When selected, the Analyst refine solution can run on a remote machine. This is a secondary option. If it does not display, click the button in the dialog box. See “Utilizing Remote Computing” for details on remote simulation.
Archive Refine Solution Files to Project - Controls whether to save in the Microwave Office project file the auxiliary files generated in an initial Analyst simulation or previous refine solution. When the auxiliary files are archived in the project, you can copy the project file to another folder or machine while preserving the ability to refine solution. Note that this can substantially increase the size of the project file. This is a secondary option. If it does not display, click the button in the dialog box.
The Analyst solver uses an Adaptive Mesh Refinement (AMR) algorithm that solves the problem multiple times with progressively finer meshes. The mesh refinement is determined by which areas of the current solution contribute the most error to the final solution. The AMR process continues until the change in the convergence metric is less than a user-defined tolerance, or the maximum specified number of AMR iterations is reached. In "Ports only" solves, the convergence metric is either Zc or Kz, depending on availability, and in "Full" solves the convergence metric is the S-parameters (magnitude or magnitude/phase). Since the number of AMR iterations affects both accuracy and solution time, the options that control AMR (particularly the convergence tolerance) are among the most important solver options. The Target Mesh Growth Fraction is universal to the simulation, but you can set the other options independently for "Ports only" solves and "Full solves".
Maximum Iterations - The maximum number of AMR iterations permitted in each phase of the simulation. If the sequence has not terminated by reaching the convergence tolerance, it terminates after this number of iterations are complete.
Minimum Iterations - The minimum number of AMR iterations permitted in each phase of the simulation, regardless of the convergence status.
Minimum Converged Iterations - The minimum number of consecutive iterations with error less than the convergence criterion, before the AMR process is terminated.
Frequency Modifier - By default, the AMR process only solves at a single frequency point, and the full frequency sweep is only done for the final step. For example, if the frequency setup specifies solving at 1, 2, 3, 4, and 5 GHz, and the Frequency Modifier is set to Mid, then the initial AMR results are performed at 3 GHz. You can choose All if you want to use all frequencies, although this slows the simulation time significantly. You can also specify any AMR frequency you want by choosing Custom.
Result Convergence/Target - Only visible for the "Full solve" part of the AMR process. Controls if AMR stops on either the magnitude of the S-parameters or both the Magnitude and Phase of the S-parameters.
Result Convergence/Tolerance - Controls the stopping criterion for the "Ports only" part of the AMR process; it is a fractional tolerance. A value of 0.01 indicates that the solution is considered converged once the convergence metric changes by less than 1% from one iteration to the next.
Result Convergence/Maximum Delta Magnitude S - Controls the stopping criterion for the "Full solve" part of the AMR process; it is a fractional tolerance. A value of 0.01 indicates that the solution is considered converged once the magnitude of the S-parameters changes by less than 1% from one iteration to the next.
Result Convergence/Maximum Delta Phase S - Only visible if the Target is Magnitude and Phase S. Controls the stopping criterion for the "Full solve" part of the AMR process; it is an angle tolerance. A value of "5 Deg" indicates that the solution is considered converged once the phase of the S-parameters changes by less than 5 degrees from one iteration to the next.
Result Convergence/Ignore Phase When Magnitude Less Than - Only visible if the Target is Magnitude and Phase S. If the magnitude of a term in the S-parameters is less than this value, the phase of this term is not considered in determining AMR convergence.
Coarsening Enabled - Controls whether the elements can be made larger in any portion of the mesh during adaptive mesh refinement.
Target Mesh Growth Fraction - Controls how rapidly the mesh grows with some exceptions related to wavelength and skin-depth based refinement. A value of 0.2 indicates that the element count generally does not increase by more than 20% from one iteration to the next.
You should choose your AMR frequency wisely. The AMR frequency and basis set choice place a cap on the initial mesh element size, to ensure that the system is properly sampled for the highest AMR frequency at the beginning of the simulation. The mesh refines at this frequency until the specified accuracy is obtained. If you are simulating a filter and your AMR frequency is outside the bandwidth of the filter, you do not get an accurate answer. There is generally no signal through the filter at this frequency, so the mesh converges quickly since there is little change in the solution. For example, consider a simple bandpass filter centered at 4.4 GHz with about 10% bandwidth. The return loss of the filter for an AMR frequency of 4.4 GHz and 5.0 GHz is shown in the following figure.
You can view the final mesh to see if it looks reasonable. In this case the 4.4 GHz mesh looks good.
The 5.0 GHz mesh does not look good.
Analyst EM documents simulate the same way as other EM structures in the software. After creating or editing a structure, you simulate it by clicking thebutton on the Standard toolbar, or by choosing . You can also right-click an Analyst EM document in the Project Browser and choose to simulate only the selected document. You are prompted for the name of the data set.
After an Analyst structure simulation finishes because the solution converges to a tolerance you specify, or the simulator completes the number of adaptive mesh refinement (AMR) iterations you set, aoption is available by right-clicking the document in the Project Browser.
If there is no change to the convergence option, this command runs one more iteration of the adaptive mesh refinement. It reuses the mesh and field solutions at the last AMR sequence, saving the time it took to get to the last AMR step.
If the Maximum Delta Magnitude S option is reduced after the simulation terminates and then you run the command, the Analyst solver performs one or more iterations of mesh refinement until the new convergence tolerance is met.
Analyst simulations start the same way as other simulations in the NI AWR Design Environment software, however unlike AXIEM, the AMR process controls the simulation. As the AMR process proceeds, results are sent back to the NI AWR Design Environment program, allowing you to monitor the AMR progress. The AMR process is broken into three phases:
This step refines the mesh until a converged port solution is obtained. The Ports Only AMR is only relevant to wave ports, but it executes and converges immediately if only lumped ports are present. The convergence criterion is either the propagation constant Kz or the characteristic impedance Zc, depending on availability. The AMR Phase 1 (Ports only) category includes options that affect this step. Typically the defaults are sufficient for this type of solution.
After the "Ports only" AMR terminates, the simulator moves into a full volumetric solve. Here the convergence criteria is based on the scattering matrix. This criteria computes the maximum change in the magnitude of S (across all AMR frequencies and all terms in S). The AMR Phase 2 (Full Solve) category includes options that affect this step. Phase 2 calculates an estimate of the memory needed for the next step. If it determines that further iterations will cause you to run out of memory, it aborts this phase with an appropriate message and moves to phase 3. You should get a result if this occurs, but the result is not fully converged, so use it with caution.
After the "Full solve" AMR terminates, the simulator runs one final solve across all frequencies using the final mesh. On the structure Options dialog box Analyst tab, the Solver category includes options that affect this step, although many of the options in this category are also used during the AMR process. The Analyst solver automatically skips this step if it is unnecessary due to the requested set of frequencies and associated Frequency Modifier option selections in the AMR categories. If every simulation frequency was used during the AMR process, the final solve is not necessary and is skipped.
The Status Window displays any messages from the solver as the simulation runs. Some of the messages also display in the Output log. Inconsistencies in the geometry or assumptions that the solver is required to make are listed in the warning messages. For all simulations, you should view any warning messages listed here to ensure that the setup and geometry are defined as intended.
In the event of simulation failure, details of the output errors are listed in the Status Window. Errors typically result from problems in the geometry that the solver cannot automatically resolve, such as improper use of ports or boundary conditions. These errors should help you determine the cause of the failure so you may correct the problem and restart the simulation.
The Analyst simulation log shows details about the current state of the simulation, as the calculations are performed. Most information on the AMR process displays in the run table, which provides an overview of the progress and current state of the simulation during the AMR sequence. The following is an example run table.
This table represents the middle of an ongoing simulation. The information included is summarized as follows:
Iteration: The iteration number and type. The text indicates whether each iteration was a "Ports only" solve, or a "Full solve".
Elements: The number of mesh elements used to find the solution. Depending on the geometry and materials, this number may be smaller than the total number of elements in the system. Elements inside metals or in electrically isolated regions of the geometry do not impact the final solution, and as a result are not used in the FEM calculation. This value indicates the number of elements that are used in the FEM calculation; so the value is less than or equal to the number of elements generated in the system.
Delta Zc/Kz (0.01): Either the characteristic port impedance (Zc) or the propagation constant (Kz) is used as a metric for convergence in the port solve, depending on the port configuration. As the solution converges, the change of Zc or Kz from one solution to the next decreases. The value in parentheses (here, 0.01) indicates the threshold for convergence required by the solver. Once Delta Zc/Kz reaches this threshold, the port solution is considered converged and an asterisk displays next to the value as shown, circled in red.
Delta S (0.001): The system S-parameters are used as a metric for convergence in the "Full" solve. During the "Ports only" solves, S is not calculated so the value in this column is "n/a". Once the solver performs the "Full" solves, the value in this column indicates the change in magnitude S (and also phase S if appropriate) from one iteration to the next. The value in parentheses (here, 0.001) indicates the threshold for convergence required by the solver. Once Delta S reaches this threshold, the full solution is considered converged and an asterisk displays next to the value. In this example, the full solution is not yet converged.
Time (min): The elapsed time since the start of the simulation, in minutes.
Peak Memory (MB): The first value indicates the maximum amount of memory in MB used so far for the simulations. The value in parentheses indicates the same quantity in terms of percent of allotted memory.
In each new iteration, the results of the previous iteration are used to refine the mesh. Details of the mesh changes are presented after the run table, showing the number of elements refined or coarsened, and giving the resulting final element count. This element count indicates the full number of elements in the mesh, including those that do not participate in the field solve.
After the AMR sequence is complete and the solver is performing the final solve, the Output log shows the progress of the final solve. The contents of the progress window during the final solve depends on the type of frequency sweep, as described in “Solver Options”. For both the discrete frequency sweep and the fast frequency sweep, each active process reports each frequency as it begins processing it. For the fast frequency sweep, this frequency corresponds to f0 in “Frequency Sweeps”.
When a simulation converges, an asterisk displays next to the convergence metric in the run table, as described in the previous section. The asterisk is a visual indicator that the convergence metric is below the threshold. In the previous example, Delta Zc/Kz is "0.00939" in Ports only 4, which is lower than the threshold of "0.01". As the solver performs "Full" solves, you can see that Delta Zc/Kz continues to decrease, indicating that convergence in the port solves is maintained throughout the mesh refinements in the "Full" solves.
The convergence threshold in the "Full" solves is set to "0.001" in this example. In the four "Full" solves shown here, Delta S reaches a value of "0.01382", which must decrease by a full order of magnitude before it is considered converged. At this rate of convergence, you can reasonably expect this simulation to require several more iterations before it converges.
This section describes using simulation data sets, viewing the Analyst structure mesh, adding 3D EM specific annotations, and controlling visibility based on the assigned EM layer.
Data Sets for Analyst simulation are mostly identical to other simulation data sets. See “Simulation Data Sets” for details on using simulation data sets. Analyst simulation is slightly different in that during the simulation phase, there is data available at each AMR sequence because each sequence produces its own sub-data set. For example, the data sets display similar to the following figure when a simulation runs or is complete.
The sub-data set node icons include a "P" to indicate a Port Only AMR sequence. After the simulation is complete, you can update the data to each sub-data set to see the graph data as well as mesh and any current of field annotations displayed on the 3D view. This is a good way to view the mesh being refined at each AMR step.
NOTE: The sub-data sets are never saved in the project because they can require significant disk space and are typically only needed directly after the simulation is complete, to see its progression. They are still available if you close and reopen the project, however, they are no longer available if you save the project and move it to a new location.
You can view the Analyst structure mesh in the EM structure 3D view. With a 3D view window active, click thebutton on the toolbar. If the structure is not previously simulated, an initial mesh is generated.
A "MESH_3D" annotation is added under the EM document in the Project Browser. Double-click the annotation to view the available options. See “Volumetric 3D Mesh: MESH_3D” for annotation details. Click the button again to turn off the mesh display. The Layout Manager Visibility by Material/Boundary pane allows you to quickly show/hide mesh elements by material type and verify that all the material properties are set up as expected.
When you view the mesh before a simulation runs, you see the initial mesh used during simulation. If you keep the 3D view open with the mesh turned on during simulation, you see the mesh update after each AMR iteration. If you view the mesh after the simulation runs, you see the final mesh for the structure.
Cut planes are a useful tool when looking at a mesh. One side of the plane shows the mesh, while the other side does not, and the plane location is easily moved. See “Cut Planes” for details.
The following are 3D EM specific annotations you can add to an Analyst document. See “EM Annotations and Cut Planes” for details on adding EM annotations.
EM_FIELD_CUT - Click the “EM Field on Cut Plane: EM_FIELD_CUT” for details.button on the toolbar to add this annotation. You can use it to visualize fields on cut planes. See
EM_FIELD_CARPET - Allows visualization of field-based carpet plots. See “EM Field Carpet Plot on Cut Plane: EM_FIELD_CARPET” for details.
EM_FIELD_CONT - Allows surface contour plots to be created. See “EM Field Surface Contours: EM_FIELD_CONT” for details.
EM_FIELD_VECT - Allows visualization of field vectors. See “EM Field Vectors: EM_FIELD_VECT” for details.
You can use the Layout Manager Visibility by EM Layer pane to control visibility based on the assigned EM layer. The Visibility by Material/Boundary pane is useful for selectively viewing boundary condition setup and mesh properties. It only changes visibility of the 3D view of the structure. The following figure shows this pane.
In this dialog box you can:
Click a column header to sort by that column's values.
Click a specific color and use the drop-down menu to change the color for that visibility.
Click the Visible column next to a material to turn the visibility for that material on or off.
This section provides details on Analyst simulator operation.
The Analyst solver is based on the finite-element method, in which the solution of Maxwell's equations is expanded on each tetrahedron in terms of basis functions. These basis functions have local support (they are zero outside the element), and they have unknown amplitude inside the element. These unknown amplitudes are determined by the finite-element method. Analyst simulation uses an electric field formulation on hierarchical vector basis functions in tetrahedrons with flat sides. Each set of basis functions is called a basis set. The basis sets in Analyst simulation are defined in terms of the accuracy of the field interpolation expected from each group of basis functions. Included within Analyst simulation are basis sets which may represent the field with up to fifth-order accuracy.
Based on historical and academic precedent, the basis set of order x+1 accuracy is designated "hx.5" within Analyst simulation, with the interpolation order for the electric field given by x+1, and the interpolation order for the magnetic field given by x. In general, as you increase x, you obtain higher accuracy in the field solution at the cost of an increased computational burden in the calculation. The first order, linear basis set consists of six functions, one associated with each edge in the tetrahedron. This basis set is designated h0.5; it represents the electric field to first-order, and the magnetic field (obtained by taking the curl of the electric field) to zero-th order. The second order basis set (h1.5, representing the electric field to second-order and the magnetic field to first-order) contains the six linear functions of the h0.5 set, as well as 14 additional functions (one for each edge and two for each face on the tetrahedron). This composition is a characteristic of hierarchical basis sets. Similarly, there are 45 functions in the cubic set h2.5, 84 in the quartic set h3.5, and 145 in the quintic set h4.5.
Because the number of functions and associated unknowns depends on the basis set, so too do the computer resources required to solve a problem using the basis set. Generally, the time needed to solve a problem increases in multiple for each increment in the basis set order. For example, if it takes X seconds to solve a problem using the h0.5 basis set on a given mesh, it generally takes 5X to 10X to solve on the mesh using h1.5, and 10X to 20X for h2.5. Direct timing comparisons are difficult as there are other factors that affect run times, such as how fast the AMR process converges, so the choice of basis set is often made on other grounds. Often the choice is determined by experimentation to discover what works best on a given class of problem. When in doubt, start with the quadratic h1.5 basis set.
Analyst simulation includes capacity for hybrid basis sets, in which the accuracy of the basis set may vary within the mesh depending on the system geometry. These hybrid sets are designated as "cx.5", and they are hierarchical vector basis sets. They are equivalent to the h0.5 basis set in regions of the model away from conducting corners and edges, but use higher orders (up to hx.5) basis sets on corners and edges. Hierarchical basis sets generally give solution accuracies similar to the corresponding hx.5 basis sets, but at a lower computational cost.
Note that, in Analyst simulation, the basis set for the "Ports only" solve is automatically one order higher than the chosen "Full" solve basis set. Since the default basis set for the "Full" solve is h1.5 (quadratic), the default for the "Ports only" solve is h2.5 (cubic).
The Analyst solver supports two kinds of frequency sweeps: (1) discrete sweep, where each frequency is solved separately, and (2) fast frequency sweep, where an asymptotic method is used to determine the solution over the frequency band using a small number of discrete solves. In a discrete sweep, each frequency is effectively treated as a separate problem, and as such, discrete sweeps make efficient use of multiple processors since solving for each frequency can proceed independently.
The fast frequency technique used in Analyst simulation is based upon the Galerkin Asymptotic Wave Expansion method (GAWE). In this approach the finite-element matrix equation is expanded about a specified frequency (expansion point f0) in terms of a power series in f – f0, where the initial choice of f0 is chosen adaptively by the solver. The solution at each fi near the expansion point is then approximated by a sum over basis vectors that are determined from the matrix equation expansion. Generally the accuracy of the expansion degrades as |fi – f0| gets larger, causing the process to yield acceptable answers only within a range fmin < f0 < fmax, where the bounds are determined by checking the residual of the original matrix equation. Converged frequencies are archived, and new expansion points are then picked in unconverted regions of the band, with the process continuing until all frequency points are converged. Processing associated with individual expansion points is independent, so when multiple processors are used each one works on a distinct expansion point, which allows for efficient parallel processing.
If you specify the sweep type as Automatic, you get either a Discrete sweep or a GAWE sweep, depending on which is likely to be faster for the given frequency count, port count, and parallel configuration. The Discrete sweep is typically faster for simulations with fewer frequencies per active process, especially if the simulation includes many ports. The GAWE sweep is typically faster for a large frequency count with a relatively small number of ports. Even with GAWE, you generally do not want to ask for more frequencies than you need, as there is an additional computational burden for each frequency point. The fast frequency sweep also requires more computer memory than the discrete sweep because of the need to store matrix expansions and basis vectors, and the amount of time it takes for the fast frequency sweep to finish is a function of the number of ports/modes, since expansions must be formed for each source in the problem.
The finite-element method used by the Analyst solver is appropriate for high frequencies; it loses accuracy at very low frequencies. To protect against simulation at frequencies that are too low, the Analyst solver defines a minimum solved frequency (the default value is 10 MHz) below which it does not apply the finite-element method. You can change the Minimum Solved Frequency in the EM Options dialog box on the Analyst tab. Adjust this value with care; results for simulations run at very low frequencies may lose accuracy. If you attempt to simulate at a frequency below the minimum solved frequency, the Analyst solver extrapolates the results from above the minimum solved frequency using a cubic polynomial fit. Previous versions of the Analyst solver simply replaced all frequencies below the minimum solved frequency with the minimum solved frequency. To obtain this behavior, set Extrapolate Toward DC to None. Since the results below the minimum solved frequency are obtained by extrapolation and not through application of electromagnetic principles, you should view the low-frequency results with some skepticism. These results are provided for convenience only and are not intended to provide true physical insight. Nevertheless, if the lowest frequency you request is close to the minimum solved frequency, the extrapolation often provides good results.
This section provides details on the higher level functionality of the Analyst simulator.
You can calculate the radiation patterns of antennas using the Analyst solver, and plot them as a 3D annotation or 2D graph by adding the appropriate antenna measurement. Antenna pattern is not calculated by default because it adds simulation time and increases the size of the simulation results stored. This section guides you through the setup to obtain antenna patterns in Analyst simulation.
Far field patterns are calculated from the fields within the EM structure. Right-click the Analyst structure for which you want to make antenna measurements and choose Options to display the Options dialog box. On the Analyst tab, under Antenna Far Fields set Output Frequency to All Frequencies to output antenna patterns at all solve frequencies, to AMR Frequencies Only to output antenna patterns at the AMR frequencies, or to None to avoid calculating far fields.
NOTE: Outputting the fields results in longer simulation times and significantly larger data sets. To minimize this effect, be cautious when choosing the frequencies of the EM structure.
The sphere sampling increment gives the spacing between field solve points on the far field sphere. By default, the sphere sampling increment for each direction (Theta and Phi) is set to 5-degrees. To override this setting, right-click the Analyst structure and choose Options to display the Options dialog box. On the Analyst tab, click the Show Secondary button and change the Sphere Sampling Increment, Theta and the Sphere Sampling Increment, Phi to the desired step.
To obtain antenna measurement results in Analyst simulations, you must apply radiation boundary conditions so that energy is not trapped within the enclosure. For best results the domain exterior should be a combination of open boundary conditions, symmetry planes, and ground planes. It is best to avoid contact between any other structure and the open boundaries, in general. For the most accurate results you should change any Approx Open boundary condition to Perfectly Matched Layer. The Analyst solver allows for using both PML and Approx Open boundary conditions in one structure for general simulations, but when calculating antenna results only one may be used in a geometry. For more information about the different boundary conditions see “Specifying Simulation Boundaries”. If a boundary is used as a ground plane for an antenna structure (for example, a patch antenna), then you can set the boundary to proper material or it can be simulated as a PEC.
You can use the Electric Symmetry boundary condition to lower the number of elements and simulation time of an antenna structure where there is natural symmetry in the fields. There are limitations on the Electric Symmetry boundary condition, such as two boundaries that are parallel to one another cannot both be set to this boundary condition. For more information on the electrical symmetry boundary condition see “Choosing Boundary Conditions”.
When you use the Analyst solver for computing far field patterns, the radiating structure should be at least a quarter wavelength away from the radiation boundary. You need to carefully set up waveport excitation when located on a boundary. A waveport assigned on an enclosure face overwrites any other boundary condition, including the Perfectly Matched Layer boundary condition. To avoid this problem you should use internal wave ports. You can also avoid this problem by creating the antenna using the Analyst 3D Editor and applying the waveport and radiation condition on different areas of a boundary face. This is illustrated in the following figure where red indicates the waveport at the end of a coaxial waveguide, and blue denotes PML boundary. In this case it is necessary to assign the waveport to the dielectric layer. Notice that this designation is impossible using the NI AWR Design Environment Layout Editor because the waveport would have to occupy the entire enclosure face.
You can also set up the enclosure so the waveport lies outside, as shown in the following figure. Here, the gold color indicates the portion of the waveguide that extrudes outside the enclosure face. Since the waveport is no longer coplanar with the PML, the waveport has been assigned to the center conductor of the coax as is preferred.
When plotting an Antenna measurement (for Measurement Type, choose Electromagnetic > Antenna) for an Analyst structure, ensure that the Sweep Freq is set to FSAMP as shown in the following figure.
NOTE: If you set up the measurement before the structure is simulated, the frequency list for FSAMP does not contain the frequencies specified for the antenna measurements. If you set up the measurement after the structure is simulated, FSAMP contains the specified field output frequencies and is initially set to the first frequency in the list. Also, for Analyst antenna measurements, you must select the Use Interpolated Data check box.
For information on a particular measurement, click thebutton on the Add/Modify Measurement dialog box.
Analyst simulation can model the loss effects of conductor surface roughness for both thick and thin metal. Surface roughness is specified as a material property in the same manner as for AXIEM. See “Loss Model for Conductor Surface Roughness” for details.