Trouble defining differential feed for pcbstack feed

9 次查看（过去 30 天）
Jeffrey Andle
Jeffrey Andle 2025-7-7
回答： Sameer 2025-7-11
I have made a stacked patch antenna in pcbstack with a single feed. The S11 and beam 3D look reasonable.
Ant = pcbStack("BoardShape", ground, "BoardThickness", eHig+eHig1+eHig2);
d1 = dielectric("air");
d1.Thickness = eHig;
d2 = dielectric("air");
d2.Thickness = eHig1;
d3 = dielectric("air");
d3.Thickness = eHig2;
Ant.Layers = {patchP2, d3, patchP1, d2, patchD, d1, ground};
Ant.FeedDiameter = 0.0001;
Ant.FeedLocations = [0, fd, 5, 7]
Ant.FeedVoltage = [1];
Ant.FeedPhase=[0];
I made the system have dual feeds for H and V with only one feed active:
Ant.FeedLocations = [0, fd, 5, 7; fd, 0, 5, 7];
Ant.FeedVoltage = [1, 0];
Ant.FeedPhase=[0, 0];
This seems to work OK. I then try to make the feeds differential (trying Y pair only first)
Ant.Layers = {patchP2, d3, patchP1, d2, patchD, d1, ground};
Ant.FeedDiameter = 0.0001;
Ant.FeedLocations = [0, fd, 5; 0, -fd, 5];
Ant.FeedVoltage = [1, 1];
Ant.FeedPhase=[0, 180]
This causes a split in the 3D pattern with +Y and -Y lobes and S11=0. The feedpoints are shown in the right locations. If I use 0 for both phases I get a proper pattern but S11 is still 0.
If I use coaxial feeds 5,7 with 0 and 180 phase it looks like twice the impedance and the more symmetric pattern I hoped to attain. I did change sparameters to 100 Ohms and at least the S11 depth is correct, so the opposing phase feeds are series connected. This makes "sense" since the single feed was matched to 50, that the differential coax probe would match to 100 until the design is modified, but why didn't the differential pair of probes work?
I further added X=+/-fd feeds with zero voltage, trying to simulate the inactive but terminated H probes. Again, things "look" OK, but the matching on the terminations is uncertain given the feed impedance issues.

请先登录，再回答此问题。

采纳的回答

Sameer
Sameer 2025-7-11
Hi @Jeffrey Andle
The issue you're seeing with differential feeding in "pcbStack" likely comes from how feed points are interpreted. When you use:
FeedLocations = [0, fd, 5; 0, -fd, 5];
This creates two independent balanced feeds on layer 5, each referenced to ground by default. When driven 180° out of phase, the simulation treats them as separate sources, not as a true differential pair which can lead to unexpected S11 (like 0) and split beam patterns.
In contrast, when you use:
FeedLocations = [0, fd, 5, 7; 0, -fd, 5, 7];
You're defining coaxial feeds from ground layer 7 to signal layer 5. This creates a proper differential excitation when used with:
FeedVoltage = [1, 1];
FeedPhase = [0, 180];
This configuration connects the two feeds through the ground, forming a series connection. That’s why S11 improves when you set the reference impedance to 100 Ω , you're effectively modeling two 50 Ω feeds in series.
Use N×4 FeedLocations (coaxial) for differential setups, not N×3 (balanced), and adjust impedance in "sparameters" accordingly.
Hope this helps!

更多回答（1 个）

Ruchika Parag
Ruchika Parag 2025-7-10
Hi @Jeffrey Andle, it looks like you're running into expected behavior when using differential feeds with pcbStack, especially with delta-gap probes.
When you define differential feeds using two delta-gap probes like this:
Ant.FeedLocations = [0, fd, 5; 0, -fd, 5];
Ant.FeedVoltage = [1, 1];
Ant.FeedPhase = [0, 180];
You're effectively creating two separate feed ports, but pcbStack doesn't automatically treat them as a single differential port. Because of that, S11 can show unexpected results (like being zero), and the beam pattern can look split or inconsistent—even though the red spheres show the feeds are placed correctly.
On the other hand, when you use coaxial feeds like:
Ant.FeedLocations = [0, fd, 5, 7; 0, -fd, 5, 7];
Ant.FeedVoltage = [1, 1];
Ant.FeedPhase = [0, 180];
you're defining balanced coaxial probes that simulate a proper differential port. This setup is treated as one combined excitation, and the impedance (e.g., 100 Ω) behaves as expected. That's why you saw a more symmetric pattern and correct S11 response in this case.
So in short: if you're looking for differential excitation and proper matching, it's better to use balanced coaxial feeds. Delta-gap probes can work visually but don't produce reliable S-parameters for differential modes unless handled carefully—like combining them manually or processing the results externally.
Hope this helps!

类别

在 Help Center 和 File Exchange 中查找有关 PCB Antenna Design and Fabrication 的更多信息

标签

产品


版本

R2024a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by