An unequal Wilkinson power divider can distribute microwave power across several outputs while assigning different power levels to different branches. That capability is useful in antenna feed networks, where elements may require a deliberate amplitude distribution rather than equal excitation. A reported X-band design demonstrates one approach: cascade equal-split stages in a binary arrangement, then adjust microstrip line widths and add isolation resistors to produce an eight-output network with symmetric power levels.
The design was simulated and fabricated for operation near 9.68 GHz. Its measured insertion-loss pattern follows the intended symmetry across corresponding output pairs, although the fabricated values differ from simulation. For engineers considering a multi-output microwave feed, the work highlights both the design logic and the practical importance of validating a circuit after fabrication.
Why use a Wilkinson power divider?
A basic T-junction can split a signal, but it has two important limitations: impedance mismatch and coupling between output ports. A Wilkinson divider addresses these issues by combining transmission-line sections with resistive isolation. Under the intended operating conditions, the network can provide matched ports and reduce interaction between its outputs.
Those properties make the Wilkinson topology useful in microwave circuits and antenna arrays. Its familiar three-port form can be adapted for unequal division, but achieving a large difference in output power can complicate the physical implementation. The required characteristic impedances may lead to narrow microstrip lines, which can be challenging to realize. Other approaches described in the design background—including defected ground structures, parallel-strip lines and grooved substrates—also involve trade-offs in construction, isolation, bandwidth or circuit size.
Building an eight-way unequal Wilkinson power divider
Rather than begin with a single complex multi-output junction, the design uses a binary topology. Three cascaded stages of three-port splitters create eight output paths. The equal-divider stages provide a starting architecture; unequal output levels are then introduced by changing microstrip widths and incorporating resistors to decouple ports.
This staged method makes the relationship between the network architecture and output distribution clear. The final arrangement is symmetric: outputs on opposite sides of the network have corresponding power levels. In the fabricated results, ports 2 and 9 form the lowest-power pair, while ports 5 and 6 form the highest-power pair.
The work used different substrate details at different design stages. The equal divider was designed in ADS with a 0.5 mm FR4 substrate specified at a relative permittivity of 3.38. The unequal divider was simulated in HFSS using a 0.5 mm Rogers 4003 substrate with a specified relative permittivity of 3.55. Keeping these design conditions distinct matters when interpreting the reported results or attempting to reproduce the layout.
Simulation and fabrication results at 9.68 GHz
HFSS simulation and fabrication measurements show the same overall output pattern at 9.68 GHz, with power levels arranged symmetrically across the eight outputs. The reported insertion losses are:
- Ports 2 and 9: -13.25 dB after fabrication; simulated values were -14.16 dB and -14.56 dB.
- Ports 3 and 8: -11.4 dB after fabrication; simulated values were -12.59 dB and -12.93 dB.
- Ports 4 and 7: -10.36 dB after fabrication; simulated values were -10.47 dB and -11.18 dB.
- Ports 5 and 6: -9.7 dB after fabrication; simulated values were -9.99 dB and -10.65 dB.
These figures are insertion-loss measurements from the input to each output, not direct power percentages. The paper also reports a return loss of approximately -16 dB and identifies the divider as suitable for feeding an array antenna. That reported matching result is relevant because input reflection can affect how effectively a feed network transfers power from its source.
Fabricated and simulated values are not identical. Differences across the output paths range from about 0.11 dB to 1.53 dB in the comparison reported. For implementation work, the practical lesson is to assess both the shape of the output distribution and the deviation from simulation. Preserving the intended symmetry is valuable, but the measured response is the more useful reference for the built circuit.
Design considerations for engineers
Account for unequal-line geometry
Changing a microstrip line’s width changes its characteristic impedance. In an unequal divider, the required impedance relationships can push some lines toward dimensions that are difficult to fabricate. Line widths and the isolation-resistor arrangement therefore need to be treated as central design variables, not as late-stage layout details.
Include discontinuities and layout effects
The design discussion includes a compensated 90-degree microstrip bend. Bends and other discontinuities can affect a microwave circuit’s electrical behavior, so their treatment belongs in the layout and simulation process. The source describes bend calculations and evaluates the divider using electromagnetic simulation, rather than relying only on an idealized circuit model.
Validate the completed network
A cascaded network accumulates the behavior of its individual stages and interconnections. Simulating the complete unequal divider and comparing its response with fabricated measurements helps reveal whether the overall power distribution, matching and isolation meet the intended design objective. Results should be interpreted at the specified operating frequency and substrate conditions.
Practical implications
The eight-output design shows how a binary cascade of Wilkinson stages can be adapted to provide unequal, symmetric power distribution for a microwave feed network. Its implementation depends on managing microstrip impedances, isolation between branches, layout discontinuities and fabrication variation. For engineers developing an antenna feed or another X-band microwave circuit, the key takeaway is to design the output ratio and physical layout together, then verify the completed network through both simulation and measurement.