Difference between revisions of "Application Note 3: Designing A Slot-Coupled Patch Antenna Array With A Corporate Feed Network Using EM.Picasso"
From Emagtech Wiki
Kazem Sabet (Talk | contribs) |
Kazem Sabet (Talk | contribs) (→Constructing a 16-Element Patch Array) |
||
| (31 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
| − | {{projectinfo| | + | {{projectinfo|Application| Designing A Slot-Coupled Patch Antenna Array With A Corporate Feed Network Using EM.Picasso|ART PATCH Fig title.png|In this project, we will build and analyze a 16-element slot-coupled patch antenna array with a microstrip corporate feed network.| |
| − | *[[CubeCAD]] | + | *[[Building Geometrical Constructions in CubeCAD | CubeCAD]] |
| + | *[[EM.Picasso]] | ||
*PEC Traces | *PEC Traces | ||
| − | * | + | *Slot Traces |
*Mesh Density | *Mesh Density | ||
*Scattering Wave Port | *Scattering Wave Port | ||
| − | * | + | *Strip Gap Circuit |
*Radiation Pattern | *Radiation Pattern | ||
|All versions|None }} | |All versions|None }} | ||
| Line 15: | Line 16: | ||
== Designing the Patch Radiating Element == | == Designing the Patch Radiating Element == | ||
| − | + | The operating frequency of the patch array is f = 2.4GHz. At this frequency, the free-space wavelength is λ<sub>0</sub> = 125mm. The patch radiators will be spaced at half free-space wavelength: S<sub>x</sub> = S<sub>y</sub> = λ<sub>0</sub>/2 = 62.5mm. The design of the slot-coupled patch antenna is described in detail in [[EM.Picasso Tutorial Lesson 7: Designing A Slot-Coupled Patch Antenna]]. The substrate consists of two finite-thickness dielectric layers with ε<sub>r</sub> = 3.38, σ = 0, separated by a perfect electric conductor (PEC) ground plane of infinite lateral extents. The table below summarizes the substrate stackup's layer hierarchy: | |
| − | + | {| class="wikitable" | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | {| class="wikitable | + | |
|- | |- | ||
! scope="col"| Substrate Object Label | ! scope="col"| Substrate Object Label | ||
| Line 50: | Line 26: | ||
! scope="col"| Thickness | ! scope="col"| Thickness | ||
|- | |- | ||
| − | + | | THS | |
| Half-Space Medium | | Half-Space Medium | ||
| Top Substrate Termination | | Top Substrate Termination | ||
| Line 56: | Line 32: | ||
| Infinite | | Infinite | ||
|- | |- | ||
| − | + | | PEC_1 | |
| PEC Trace | | PEC Trace | ||
| Patch Plane | | Patch Plane | ||
| Line 62: | Line 38: | ||
| 0 | | 0 | ||
|- | |- | ||
| − | + | | Layer_1 | |
| Substrate Layer | | Substrate Layer | ||
| Patch Substrate | | Patch Substrate | ||
| − | | ROGER | + | | ROGER RO4003C |
| 2mm | | 2mm | ||
|- | |- | ||
| − | + | | PMC_1 | |
| − | | | + | | Slot Trace |
| Slot Plane | | Slot Plane | ||
| PMC | | PMC | ||
| 0 | | 0 | ||
|- | |- | ||
| − | + | | Layer_2 | |
| Substrate Layer | | Substrate Layer | ||
| Feed Substrate | | Feed Substrate | ||
| − | | ROGER | + | | ROGER RO4003C |
| 0.787mm | | 0.787mm | ||
|- | |- | ||
| − | + | | PEC_2 | |
| PEC Trace | | PEC Trace | ||
| Microstrip Feed Plane | | Microstrip Feed Plane | ||
| Line 86: | Line 62: | ||
| 0 | | 0 | ||
|- | |- | ||
| − | + | | BHS | |
| Half-Space Medium | | Half-Space Medium | ||
| Bottom Substrate Termination | | Bottom Substrate Termination | ||
| Line 94: | Line 70: | ||
|} | |} | ||
| − | + | The design variables in this problem include the side dimensions of the square patch radiator, length and width of the coupling slot and the length of the open microstrip stub extended beyond the coupling slot. The width of the mircostrip feed line is chosen to be w<sub>f</sub> = 2.4mm to yield a characteristic impedance of Z<sub>0</sub> = 50Ω. | |
| − | + | {| class="wikitable" | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | {| | + | |
|- | |- | ||
| − | + | ! scope="col"| Design Variable Name | |
| − | + | ! scope="col"| Optimal value | |
| − | + | ||
|- | |- | ||
| − | + | | patch_len | |
| − | + | | 39.5mm | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| − | + | | slot_len | |
| − | | | + | | 12mm |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| − | + | | slot_wid | |
| − | | | + | | 1.5mm |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| + | | stub_len | ||
| + | | 21mm | ||
|} | |} | ||
| + | == Designing the Wilkinson Power Divider == | ||
| − | + | The input signal power must be divided equally among 16 patch radiating elements. In other words, a 1:16 power distribution network is needed for this project. The design of a Wilkinson power divider is described in detail in [[EM.Picasso Tutorial Lesson 9: Designing a Microstrip Wilkinson Power Divider]]. An Ω-shaped microstrip ring is used to create a three-port network. The input and output microstrip lines all have a width of 2.4mm with Z<sub>0</sub> = 50Ω. The microstrip partial ring has a width of √2Z<sub>0</sub> = 70.7Ω and serves as the two quarter-wave arms of the Wilkinson power divider. It is determined that if a lumped 100Ω resistor is connected between the two output arms of this divider, better return loss and isolation levels are achieved. The figure below shows the geometry of the optimized 1:2 Wilkinson power divider. | |
| − | + | <table> | |
| − | + | <tr> | |
| − | + | <td> | |
| − | + | [[Image:Picasso L9 Fig14.png|thumb|left|480px|The geometry of the Wilkinson power divider with the lumped resistor.]] | |
| − | + | </td> | |
| − | + | </tr> | |
| − | + | </table> | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | | | + | |
| − | + | ||
| − | | | + | |
| − | | | + | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| + | == Constructing a Four-Element Patch Sub-Array == | ||
| − | + | A binary H-tree structure is used to construct a 1:4 Wilkinson power divider network as shown in the figures below. In this case, the network involves three ring-type Wilkinson power dividers. | |
| + | |||
| + | <table> | ||
| + | <tr> | ||
| + | <td> | ||
| + | [[Image:ART PATCH Fig1.png|thumb|left|640px|The geometry of the four-element slot-coupled patch sub-array with a corporate feed network.]] | ||
| + | </td> | ||
| + | </tr> | ||
| + | </table> | ||
| − | + | <table> | |
| − | + | <tr> | |
| − | + | <td> | |
| − | + | [[Image:ART PATCH Fig2.png|thumb|left|640px|The geometry of the four-element slot-coupled patch sub-array with the patches in the freeze state.]] | |
| − | + | </td> | |
| − | + | </tr> | |
| − | + | </table> | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | | | + | |
| − | | | + | |
| − | | | + | |
| − | |- | + | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| + | The multilayer structure is parameterized with the design variables listed in the table below. Of these variables, only the open stub length needs to be changed to 18.5mm, and rest of them retain their original value for the best input impedance match. | ||
| − | + | {| class="wikitable" | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | {| | + | |
|- | |- | ||
| − | + | ! scope="col"| Design Variable Name | |
| − | + | ! scope="col"| Optimal value | |
| − | + | ||
|- | |- | ||
| − | + | | patch_len | |
| − | + | | 39.5mm | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| − | + | | slot_len | |
| − | | | + | | 12mm |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| − | + | | slot_wid | |
| − | + | | 1.5mm | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | | 1 | + | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| − | + | | stub_len | |
| − | | | + | | 18.5mm |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
|- | |- | ||
| + | | resistance | ||
| + | | 100 Ohms | ||
|} | |} | ||
| + | The figure below shows the planar mesh of the sub-array. The patch and slot elements are discretized with a mesh density of 30 cells per effective wavelength, while the corporate feed network requires a higher mesh density of 50 cells per effective wavelength due to the narrow line hosting the lumped resistors. | ||
<table> | <table> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig3.png|thumb|left|640px|The hybrid planar mesh of the four-element slot-coupled patch sub-array with a corporate feed network.]] |
</td> | </td> | ||
</tr> | </tr> | ||
</table> | </table> | ||
| − | + | The 4-element slot-coupled patch sub-array is simulated using [[EM.Picasso]]'s planar method of moments (MoM) solver. An adaptive frequency sweep is performed to compute the frequency response of the structure over the frequency range [2.2GHz - 2.6GHz]. The figures below show the variation of the sub-array's return loss with frequency and its 3D far-field radiation pattern computed at 2.4GHz. | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | [[ | + | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
<table> | <table> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig4.png|thumb|left|480px|The return loss of the 4-element patch sub-array over the frequency range [2.2GHz - 2.6GHz].]] |
</td> | </td> | ||
</tr> | </tr> | ||
| Line 535: | Line 168: | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig5.png|thumb|left|640px|3D radiation pattern of the 4-element patch sub-array computed at 2.4GHz.]] |
| − | + | ||
| − | + | ||
| − | + | ||
</td> | </td> | ||
</tr> | </tr> | ||
</table> | </table> | ||
| − | == | + | == Constructing a 16-Element Patch Array == |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| + | The binary H-tree structure described earlier is expanded to construct a 1:16 Wilkinson power divider network as shown in the figures below. In this case, the network involves 15 ring-type Wilkinson power dividers. | ||
<table> | <table> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig10.png|thumb|left|640px|The geometry of the 16-element slot-coupled patch array with a corporate feed network.]] |
</td> | </td> | ||
</tr> | </tr> | ||
| Line 569: | Line 188: | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig11.png|thumb|left|640px|The geometry of the 16-element slot-coupled patch array with the patches in the freeze state.]] |
</td> | </td> | ||
| + | </tr> | ||
| + | </table> | ||
| + | |||
| + | Using the same mesh densities as before, the planar mesh shown in the figure below is generated for the 16-element patch array. | ||
| + | |||
| + | <table> | ||
| + | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig12.png|thumb|left|640px|The hybrid planar mesh of the 16-element slot-coupled patch array with a corporate feed network.]] |
</td> | </td> | ||
</tr> | </tr> | ||
</table> | </table> | ||
| − | + | The matrix size for this planar MoM simulation is N = 10,771. [[EM.Picasso]]'s LU solver was used to solver the linear system. The total computation time including the LU decomposition, back-substitution and computation of the full 3D far-field radiation pattern at an angular resolution of 1° along both the azimuth and elevation directions was 150 seconds. At the end of the planar MoM simulation, the following port characteristics are reported: | |
| + | S11: 0.447781 + 0.118984j | ||
| + | |||
| + | S11(dB): -6.682387 | ||
| + | |||
| + | Z11: 123.053609 + 37.286922j | ||
| + | |||
| + | Y11: 0.007443 - 0.002255j | ||
| + | |||
| + | The figures below show the 3D far-field radiation pattern as well as 2D Cartesian radiation pattern cuts in the principal YZ and ZX planes computed at 2.4GHz. A directivity of D<sub>0</sub> = 17.3dB is predicted for this array. | ||
<table> | <table> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig13.png|thumb|left|640px|3D far-field radiation pattern of the 16-element patch array computed at 2.4GHz.]] |
</td> | </td> | ||
</tr> | </tr> | ||
| Line 591: | Line 226: | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig14.png|thumb|left|480px|The 2D Cartesian radiation pattern of the 16-element patch array in the YZ principal plane.]] |
| − | + | ||
| − | + | ||
| − | + | ||
</td> | </td> | ||
</tr> | </tr> | ||
| + | </table> | ||
| + | |||
| + | <table> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig15.png|thumb|left|480px|The 2D Cartesian radiation pattern of the 16-element patch array in the ZX principal plane.]] |
| − | + | ||
| − | + | ||
| − | + | ||
</td> | </td> | ||
</tr> | </tr> | ||
</table> | </table> | ||
| − | + | The figures below show the surface electric current distribution maps on the patch and feed planes, as well as the surface magnetic current distribution map on the middle ground plane, all computed at 2.4GHz. | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
| − | + | ||
<table> | <table> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig16.png|thumb|left|640px|The surface electric current distribution map on the feed network plane at 2.4GHz.]] |
</td> | </td> | ||
| + | </tr> | ||
| + | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig17.png|thumb|left|640px|The surface electric current distribution map on the patch radiators at 2.4GHz.]] |
</td> | </td> | ||
</tr> | </tr> | ||
<tr> | <tr> | ||
<td> | <td> | ||
| − | [[Image: | + | [[Image:ART PATCH Fig18.png|thumb|left|640px|The surface magnetic current distribution map on the coupling slots at 2.4GHz.]] |
| − | + | ||
| − | + | ||
| − | + | ||
</td> | </td> | ||
</tr> | </tr> | ||
</table> | </table> | ||
| + | <br /> | ||
| + | <hr> | ||
| + | [[Image:Top_icon.png|30px]] '''[[#Introduction | Back to the Top of the Page]]''' | ||
| + | [[Image:Back_icon.png|30px]] '''[[EM.Cube#EM.Cube Articles & Notes | Check out more Articles & Notes]]''' | ||
| − | + | [[Image:Back_icon.png|30px]] '''[[EM.Cube | Back to EM.Cube Main Page]]''' | |
| − | + | ||
| − | + | ||
| − | + | ||
| − | [[EM.Cube | + | |
Latest revision as of 18:15, 18 May 2017
Contents
Introduction
EM.Picasso can be used to analyze large and fairly complex multilayer planar structures. In this application note, we will show how to use EM.Picasso to design a 4 × 4 slot-coupled patch antenna array with a microstrip corporate feed network. The design process involves three steps: design of the slot-couple patch element, design of the power divider, and finally, construction of the 16-element array. The first two steps are the subject of two of EM.Picasso's tutorial lessons.
Designing the Patch Radiating Element
The operating frequency of the patch array is f = 2.4GHz. At this frequency, the free-space wavelength is λ0 = 125mm. The patch radiators will be spaced at half free-space wavelength: Sx = Sy = λ0/2 = 62.5mm. The design of the slot-coupled patch antenna is described in detail in EM.Picasso Tutorial Lesson 7: Designing A Slot-Coupled Patch Antenna. The substrate consists of two finite-thickness dielectric layers with εr = 3.38, σ = 0, separated by a perfect electric conductor (PEC) ground plane of infinite lateral extents. The table below summarizes the substrate stackup's layer hierarchy:
| Substrate Object Label | Substrate Object Type | Function | Material | Thickness |
|---|---|---|---|---|
| THS | Half-Space Medium | Top Substrate Termination | Vacuum | Infinite |
| PEC_1 | PEC Trace | Patch Plane | PEC | 0 |
| Layer_1 | Substrate Layer | Patch Substrate | ROGER RO4003C | 2mm |
| PMC_1 | Slot Trace | Slot Plane | PMC | 0 |
| Layer_2 | Substrate Layer | Feed Substrate | ROGER RO4003C | 0.787mm |
| PEC_2 | PEC Trace | Microstrip Feed Plane | PEC | 0 |
| BHS | Half-Space Medium | Bottom Substrate Termination | Vacuum | Infinite |
The design variables in this problem include the side dimensions of the square patch radiator, length and width of the coupling slot and the length of the open microstrip stub extended beyond the coupling slot. The width of the mircostrip feed line is chosen to be wf = 2.4mm to yield a characteristic impedance of Z0 = 50Ω.
| Design Variable Name | Optimal value |
|---|---|
| patch_len | 39.5mm |
| slot_len | 12mm |
| slot_wid | 1.5mm |
| stub_len | 21mm |
Designing the Wilkinson Power Divider
The input signal power must be divided equally among 16 patch radiating elements. In other words, a 1:16 power distribution network is needed for this project. The design of a Wilkinson power divider is described in detail in EM.Picasso Tutorial Lesson 9: Designing a Microstrip Wilkinson Power Divider. An Ω-shaped microstrip ring is used to create a three-port network. The input and output microstrip lines all have a width of 2.4mm with Z0 = 50Ω. The microstrip partial ring has a width of √2Z0 = 70.7Ω and serves as the two quarter-wave arms of the Wilkinson power divider. It is determined that if a lumped 100Ω resistor is connected between the two output arms of this divider, better return loss and isolation levels are achieved. The figure below shows the geometry of the optimized 1:2 Wilkinson power divider.
 |
Constructing a Four-Element Patch Sub-Array
A binary H-tree structure is used to construct a 1:4 Wilkinson power divider network as shown in the figures below. In this case, the network involves three ring-type Wilkinson power dividers.
 |
 |
The multilayer structure is parameterized with the design variables listed in the table below. Of these variables, only the open stub length needs to be changed to 18.5mm, and rest of them retain their original value for the best input impedance match.
| Design Variable Name | Optimal value |
|---|---|
| patch_len | 39.5mm |
| slot_len | 12mm |
| slot_wid | 1.5mm |
| stub_len | 18.5mm |
| resistance | 100 Ohms |
The figure below shows the planar mesh of the sub-array. The patch and slot elements are discretized with a mesh density of 30 cells per effective wavelength, while the corporate feed network requires a higher mesh density of 50 cells per effective wavelength due to the narrow line hosting the lumped resistors.
 |
The 4-element slot-coupled patch sub-array is simulated using EM.Picasso's planar method of moments (MoM) solver. An adaptive frequency sweep is performed to compute the frequency response of the structure over the frequency range [2.2GHz - 2.6GHz]. The figures below show the variation of the sub-array's return loss with frequency and its 3D far-field radiation pattern computed at 2.4GHz.
 |
 |
Constructing a 16-Element Patch Array
The binary H-tree structure described earlier is expanded to construct a 1:16 Wilkinson power divider network as shown in the figures below. In this case, the network involves 15 ring-type Wilkinson power dividers.
 |
 |
Using the same mesh densities as before, the planar mesh shown in the figure below is generated for the 16-element patch array.
 |
The matrix size for this planar MoM simulation is N = 10,771. EM.Picasso's LU solver was used to solver the linear system. The total computation time including the LU decomposition, back-substitution and computation of the full 3D far-field radiation pattern at an angular resolution of 1° along both the azimuth and elevation directions was 150 seconds. At the end of the planar MoM simulation, the following port characteristics are reported:
S11: 0.447781 + 0.118984j
S11(dB): -6.682387
Z11: 123.053609 + 37.286922j
Y11: 0.007443 - 0.002255j
The figures below show the 3D far-field radiation pattern as well as 2D Cartesian radiation pattern cuts in the principal YZ and ZX planes computed at 2.4GHz. A directivity of D0 = 17.3dB is predicted for this array.
 |
 |
 |
The figures below show the surface electric current distribution maps on the patch and feed planes, as well as the surface magnetic current distribution map on the middle ground plane, all computed at 2.4GHz.
 |
 |
 |
