(Infinite Array)
2015.0 Release
• If you don’t have the project create the patch using the following dimensions and materials
Ground Plane
Radome
Thickness = 22.9mil
er = 4.1
Waveguide Plug
Radius = 284mil
Height = 251.3mil
er = 1.9
Waveguide
Radius = 284mil
Height = 641.3mil
er = 2.1
4 © 2015 ANSYS, Inc. May 6, 2015 Release 2015.0
Understanding the Lattice
Triangle Side
Note: The waveguide elements are placed on an equilateral
triangular lattice; named because an equilateral triangle can be
drawn connecting any 3 adjacent element. The distance between
adjacent elements (any side of the triangle) is 635mils. The height
of the triangle (the distance between rows in the X-Direction) is
Triangle Height 549.926mils and each row is shifted in the Y-Direction 317.5mils
with respect to the rows above and below it.
• Observe the results in the Measure Data Window
Note: The purpose of this Workshop is to model this array as an
infinitely periodic structure using the Unit Cell techniques available
in HFSS. These techniques use Master and Slave boundary pairs to
enforce a periodicity in the fields that would exist from this
infinitely periodic structure while only physically solving a single
element. It can significantly reduce the RAM and Solve Time
required to evaluate these models while taking into account all the
element to element mutual coupling. However, it does not account
for array’s edge affect which result from the nature of a finite array.
To model these affects the array must be modeled entirely in HFSS
by drawing the entire array or using Finite Array Domain
Decomposition.
6 © 2015 ANSYS, Inc. May 6, 2015 Release 2015.0
Creating the Unit Cell Design
Note: The XY cross-section of the unit cell represents the periodicity of the
array during the simulation. A proper unit cell will leave no space between
adjacent unit cells when duplicated in accordance with the array’s lattice. To
achieve this requirement, only certain shapes can be used for the unit cell.
Typical shapes include the Hexagon, Parallelogram and a modified Rectangle.
𝑨 𝟔𝟑𝟓𝒎𝒊𝒍𝒔
𝑽𝒙 = = = 𝟏𝟖𝟑. 𝟑𝟎𝟖𝟕𝒎𝒊𝒍𝒔
𝟐 𝟑 𝟐 𝟑
𝑨 𝟔𝟑𝟓𝒎𝒊𝒍𝒔
𝑽𝒚 = = = 𝟑𝟏𝟕. 𝟓𝒎𝒊𝒍𝒔
𝟐 𝟐
In the above equations |A| is the distance between any 2 adjacent elements.
These equations generally hold for an equilateral triangular lattice. The only
variation might be if one of lattice vectors lined up with the X-axis instead of
the Y-axis. In this case the X and Y component would be flipped (newVx = Vy
and newVy = Vx).
Note: The height of the unit cell was chosen to minimize the number of
modes defined on the Floquet Port. This will be discussed later in the
Workshop
• Note
• Master Boundary Window
– Name: Master1
– U Vector: New Vector…
• When you click on the Vertex, some times an error message “Lines can only
be created on selected faces” might pop-up.
• To avoid this, change the movement mode to 3D and draw the U vector
Modeler > Movement Mode > 3D
• From HFSS’s Menu Bar select HFSS > Boundaries > Assign > Master…
• Master Boundary Window
– Name: Master2
– U Vector: New Vector…
• Left click on the vertex in the upper, right corner of the face
• Left click on the vertex in the upper, left corner of the face
– Click Ok button
• From HFSS’s Menu Bar select HFSS > Boundaries > Assign > Slave…
• Master Boundary Window
– Name: Slave2
– Master Boundary: Master2
– U Vector: New Vector…
• Left click on the vertex in the upper, right corner of the face
• Left click on the vertex in the upper, left corner of the face
– Click Next button
• Phase Delay Window
– Phi: Phi_Scan
– Theta: Theta_Scan
• Click Finish button
• From HFSS’s Menu Bar select HFSS > Boundaries > Assign > Master…
• Master Boundary Window
– Name: Master3
– U Vector: New Vector…
• Left click on the vertex in the upper, right corner of the face
• Left click on the vertex in the upper, left corner of the face
– Click Ok button
• From HFSS’s Menu Bar select HFSS > Boundaries > Assign > Slave…
• Master Boundary Window
– Name: Slave3
– Master Boundary: Master3
– U Vector: New Vector…
• Left click on the vertex in the upper, right corner of the face
• Left click on the vertex in the upper, left corner of the face
– Click Next button
• Phase Delay Window
– Phi: Phi_Scan
– Theta: Theta_Scan
• Click Finish button
Note: The 01_UnitCell design should have 3 pairs of Master / Slave boundaries totaling 6 boundaries all together. Each pair should be
defined on opposite sides of the unit cell. Each boundary should have a UV coordinate system defined on it in such a way that the U-
axes point in the same direction and the V-axes point in the same direction. The only difference being a linear translation of their
origins across the unit cell.
In addition, each slave boundary should have using the Scan Angle method for defining the phase delay between the fields on the
Master and Slave boundaries. This scan angle should be controlled by two variables; Phi_Scan and Theta_Scan.
The energy associated with any mode not defined in the Floquet Port will be short circuited back into the model so its important to
include all propagating mode. However, which modes propagate depends on the angle the array is scanned. The Mode’s Calculator was
added to make choosing the appropriate modes easier. It will search a scan volume defined by Theta and Phi angles and report the
modes with the smallest attenuation constant over that scan volume. Make sure any mode not included in the port experiences at least
50dB of attenuation as it propagates from the element to the port. That way it has little chance to affect the predicted performance. The
attenuation is calculated in dB/length so the total attenuation can be computed by multiplying the calculated attenuation by the unit cell
height (0.04dB/length)*1250mils = 50dB of attenuation for mode 7. All modes listed above mode 7 have 0dB/length of attenuation so
they must be included.
Its important to note that the height of the Unit Cell impacts how many modes need to be included in the Floquet Port Setup. Non-
propagating modes can be removed from the setup by choosing a height large enough to make sure they experience the recommended
attenuation for elimination.
It’s a good idea to play around with the modes calculator a little to get a feel for how the mode attenuations are affected by frequency
and scan volume. You should see that more modes are required at higher frequencies and larger scan volumes. Therefore it is important
to specify the highest frequency you expect to simulation and the complete scan volume you expect cover. For this example that
corresponds to 10.8GHz and a scan volume extending 60deg from boresite between phi = 0deg and 90deg planes.
• Save Project
• Select the menu item File > Save
• Model Validation
• Select the menu item HFSS > Validation Check
Validate Analyze All
• Click the Close button
Note: To view any errors or warning messages, look at the Message Manager window.
• Analyze
• Select the menu item HFSS > Analyze All
• After analysis is complete save the project
• Select the menu item File > Save
Analyze All
Solution Data
• Show WavePort1 : Mode 2 Field Distribution Note: Mode 1 is a TE11 mode polarized in the
• From the Project Manager Left Click on the Y direction and Mode 2 is a TE11 mode
POEW_WS > 01_UnitCell > Port Field Display > WavePort1 > Mode 2 branch polarized in the X direction. These modes were
determined by the port Eigen Mode solution
Mode 1 Field Mode 2 Field and represent the first 2 propagating modes of
the waveguide.
• Setup the excitations for Circular Polarization Note: Since WavePort1’s modes (degenerate
• From HFSS’s menu bar select HFSS > Fields > Edit Sources… TE11 modes) are orthogonally polarized waves
• WavePort1:1: propagating down the waveguide, they can be
– Magnitude: 1 Watt used to excite a circularly polarized radiation
pattern. This is accomplished by exciting both
– Phase: 0 Deg
modes with equal magnitudes but phased +/-
• WavePort 1:2: 90 deg out of phase from each other.
– Magnitude: 1 Watt
– Phase: 90 Deg
• Click OK button
The Active Return losses are very similar to each other since
the isolation between these modes is better than -60dB.
Note: Every Optimetrics License comes with 2 DSO (Distributed Solve Option) licenses allowing a user to
run 2 instances of a parametric sweep at the same time. This speeds up the analysis of the scan volume
sweep. The number of parametric instances run simultaneously and the number of cores each instance
uses is controlled by setting up an Analysis Configuration in the HPC and Analysis Options window.
This example assumes that the computer running the analysis has 4 cores and there are 4 DSO licenses
available. The number of simultaneous analysis is determined by the Number of Tasks field. The Total
Cores field allows for each parametric instance to take advantage of multiple cores. These settings will run
for scan angle analyses at the same time with each analysis using a single core.
If the computer running this sweep does not contain 4 cores or you do not have 4 DSO licenses you will
need to reduce then number of tasks and Cores appropriately. If you have more cores and licenses
available you can further speed up the analysis by adding more tasks.
32 © 2015 ANSYS, Inc. May 6, 2015 Release 2015.0
Running Parametric Analysis
Note: This plot shows the Active Return Loss vs. Frequency for each scan angle solved during the parametric sweep. Multiple
curves were created because the Theta_Scan Value was set to All (more than one value) on the Families Tab. The plot shows that
each scan angle produces a unique impedance that gets worse as the frequency and scan angle increases.
Note: The Embedded (Active) Element Pattern can be extracted from the radiation patterns just plotted by
taking the one point on each curve corresponding to the scan angle used to produce that curve. For instance,
the Embedded Element Pattern for the angle (theta, phi) = (0deg, 0deg) would correspond to the (theta, phi) =
(0deg, 0deg) point on the RealizedGainTotal curve associated with Theta_Scan = 0deg and Phi_Scan = 0deg. All
the other points on the curve would be discarded. The points that are part of the Embedded Element Pattern
are highlighted with blue boxes in the above data table.
Note: Scan_CS is a relative coordinate system parameterized so the z-axis always points in the scan direction defined by
Theta_Scan and Phi_Scan. You can visibly verify this by switching to the Scan_CS coordinate system, changing the value for
Theta_Scan to 60deg and observing where the z-axis points. Setting up the infinite sphere using this coordinate system allows
us to plot the Embedded Element Pattern while keeping theta and phi at a constant 0deg. This makes is possible to define
the trace within HFSS. However, this only works for patterns using the Total, RHCP, LHCP, Phi and Theta polarizations. X, Y
and Z polarizations of the Scan_CS coordinate system change with respect to the global coordinate system as the beam is
steered producing incorrect results.
To fill in the rest of the pattern additional simulations would need to be performed at different scan angles.
41 © 2015 ANSYS, Inc. May 6, 2015 Release 2015.0
Embedded Element Pattern From Floquet
1) Add name
2) Insert Expression
3) Click Add
1 3
2
PowerIn = 2
Wavelength_mil = 1.18e13/10.8e9
Transmission_Coefficient =
(mag(S(FloquetPort1:1,WavePort1:1)*1*exp(cmplx(0,1)*0*pi/180)+S(FloquetPort
1:1,WavePort1:2)*1*exp(cmplx(0,1)*90*pi/180))^2+
mag(S(FloquetPort1:2,WavePort1:1)*1*exp(cmplx(0,1)*0*pi/180)+S(FloquetPort
1:2,WavePort1:2)*1*exp(cmplx(0,1)*90*pi/180))^2)/PowerIn
Gain_Floquet = 4*Pi*349203.08347/wavelength_mil^2*Transmission_Coefficient*cos(Theta_scan)
Note: These equations are based off the well established calculation
4𝜋𝐴𝑟𝑒𝑎 2
4𝜋𝐴𝑟𝑒𝑎 2
𝐺= 1− Γ 𝑐𝑜𝑠 𝑇ℎ𝑒𝑡𝑎_𝑆𝑐𝑎𝑛 = 𝑇 𝑐𝑜𝑠 𝑇ℎ𝑒𝑡𝑎_𝑆𝑐𝑎𝑛
𝜆2 𝜆2
where |T|2 is the transmission coefficient (a.k.a. transmitted Power / incident power).