Wilkinson Dividers for RX: Isolation, Loss and Alternatives
Wilkinson Dividers for RX: Isolation, Loss and Alternatives
A Wilkinson can be an excellent way to feed two receivers from one antenna, but “excellent” is conditional. Choose the topology from bandwidth, impedance, isolation, loss, balance, load behaviour and the measurements your receive system actually needs.
RF.Guru working definition: Common-mode current is the non-cancelling phasor-sum current in a specified set of conductors, evaluated at a defined cross-section and using a declared current-direction convention. In the intended differential transmission-line mode, the outgoing and return currents are equal and opposite, so their phasor sum is zero. When they do not cancel, the remaining current must close through another reference or return path—such as the outside of a coax shield, a mast, equipment chassis, station wiring, nearby structures, earth, the operator, or distributed coupling through the environment.
This broader working definition is especially useful in practical antenna systems. On transmit, non-cancelling current on the outside of the coax can make the feedline and connected structures part of the radiating antenna system unless that path is intentional, clearly defined and properly controlled—for example by providing the required return path and placing a suitable common-mode choke at the correct boundary.
When two receivers need the same antenna signal, I want more than two connectors joined together. I want a declared impedance at every port, enough receiver-to-receiver isolation, known amplitude and phase tracking, and a loss budget that still leaves both receivers externally noise limited. A Wilkinson is one strong answer—not the only answer.
Division Loss Is Not the Same as Dissipative Loss
For an equal two-way divider with matched ports, the available input power is shared between two outputs. Each output is therefore 10 log10(2) = 3.0103 dB below the total input power even in an ideal lossless divider. That is division loss: the other half of the power is delivered to the other output.
Real conductors, dielectric, transformers, components and mismatch add excess insertion loss. A useful data sheet or measurement keeps the two ideas separate. The official Keysight 87303C divider specification, for example, states insertion loss in addition to the nominal 3 dB coupling loss and separately specifies isolation, return loss, amplitude tracking and phase tracking.
The distinction matters for receive noise. Ideal division sends the incident wanted signal and antenna noise into both branches in the same power ratio. It does not by itself impose a 3 dB SNR penalty at each output. Receiver noise is then added independently in each branch. Excess passive loss before the first low-noise stage does degrade the available noise performance, and too little post-divider external-noise margin lets receiver noise reduce SNR.
What the Classic Wilkinson Promises
E. J. Wilkinson’s foundational 1960 paper describes an equiphase, equiamplitude divider with approximately matched terminals and isolation between the output ports over its design region. For the familiar equal two-way, single-section form at its centre frequency in a system impedance Z0:
- each quarter-wave branch has characteristic impedance
√2 Z0; - the resistor between the two output ports is
2Z0; - the input and both outputs are ideally matched;
- the outputs are ideally isolated; and
- the two output signals have equal amplitude and phase.
With equal matched output loads, the two output voltages are equal and the isolation resistor has no differential voltage, so an ideal resistor dissipates no power in that even-mode condition. When the output loads differ, reflected energy creates an odd-mode component and the resistor helps absorb it. Isolation, input match, output level and resistor dissipation are therefore conditions of the network and its terminations—not unconditional properties printed on the box.
Mini-Circuits AN10-006 derives the equal split and shows how mismatch changes isolation, insertion loss and internal-resistor dissipation. It also makes the measurement boundary explicit: isolation is measured between output ports while the common port is terminated correctly.
Bandwidth Comes from the Real Implementation
A single quarter-wave section is centred on one frequency and is not a 1–30 MHz universal answer. Multisection transmission-line designs can widen the useful band at the cost of size and complexity. Lumped networks, transformer hybrids and integrated implementations can cover other ratios and frequency spans, but their parasitics, core properties, winding symmetry and compensation define what “Wilkinson-like” means in practice.
Do not choose by topology label alone. Demand sweeps of all relevant S-parameters across the complete band, temperature and intended port impedances. A nominally broadband splitter can still show poor low-end match, high-end phase error or an isolation dip exactly where a strong broadcaster sits.
Isolation Is Finite and Load Dependent
Receiver inputs are not always perfect resistive terminations. Filters, protection networks, switched preamplifiers, attenuators, ADC inputs and local-oscillator leakage can make the impedance and reverse signal change with frequency and receiver state. A divider’s output isolation reduces that interaction; it does not abolish it.
When a receiver port reflects energy, some is dissipated in the isolation network, some can return toward the antenna port and some may reach the other receiver. The exact split follows the measured three-port network and all terminations. This is why one isolation number at one frequency cannot prove freedom from digital hash, LO leakage, birdies, cross-modulation or reciprocal mixing.
Also keep mode conversion separate from port isolation. A Wilkinson implemented on coaxial or planar transmission lines does not automatically provide galvanic isolation or block common-mode current on cable exteriors. If conducted common mode is part of the problem, specify the shield, enclosure, bonding and common-mode network separately and measure them.
Compare the Alternatives by the Same Requirements
| Topology | Useful when | Main trade-off | What to verify |
|---|---|---|---|
| Direct T or Y | A quick diagnostic with known high-impedance or specially designed loads | Usually poor match and receiver-to-receiver isolation in a conventional matched RF system | Actual parallel impedance, reflections, crosstalk and stability in every receiver state |
| Three-resistor matched splitter | Very wide bandwidth, DC continuity or simplicity matters more than loss | A common equal two-way form is nominally about 6 dB from input to each output: 3 dB division plus about 3 dB dissipation | Topology-specific match, isolation, resistor noise, excess loss and external-noise margin |
| Wilkinson or isolation-resistor hybrid | Low excess loss, equal in-phase outputs and useful port isolation are required over a defined band | Bandwidth and isolation depend on implementation and terminations; mismatch drives the isolation network | All three return losses, both paths, output isolation, balance and resistor limits |
| Transformer splitter | HF or wideband work where a suitable flux-coupled network offers low loss and practical size | A simple transformer split need not isolate the output ports; core, winding and parasitic behaviour limit the band | Topology, impedance transformation, isolation, balance, common mode, loss, linearity and temperature |
| Diplexer or multiplexer | Receivers need different non-overlapping bands | It separates bands rather than making equal copies of the same spectrum | Passband loss/match, stopband isolation, crossover behaviour and high-level signals |
| Active multicoupler | Many outputs need gain, filtering, switching or high isolation | Adds noise, nonlinearity, overload, reverse isolation, stability, power and failure-mode questions | Noise and gain at the real source, blocker tests, output isolation, compression, IMD and fail-safe behaviour |
The familiar 6 dB figure applies to a particular matched resistive two-way network, not to every circuit described as a resistive splitter. Likewise, “transformer splitter” does not state port isolation, phase balance or common-mode behaviour. The schematic and measured multiport data decide.
Noise Budget Before Topology
Start at the antenna or active-antenna output with available wanted power, available external-noise power and complex source impedance. Then include cable, filter and splitter loss; each receiver’s input match, noise contribution and gain state; and the strongest blockers that can reach the system.
A passive matched loss at physical temperature has a noise factor equal to its linear loss when it precedes the first active stage under the standard noise-figure boundary. But a divider used with an antenna whose external noise is already far above receiver noise needs a complete cascade calculation: the critical question is whether the signal and external noise arriving at each receiver still exceed that receiver’s added noise by the required margin.
More gain before the divider can restore margin against downstream receiver noise, but it can also overload on broadcast or nearby transmit signals and create intermodulation in every branch. A low-noise figure does not excuse inadequate P1dB, IP2/IP3, blocking, filtering or ADC headroom. The Keysight noise-figure measurement documentation distinguishes available and delivered noise power, source match and receiver contribution—exactly the distinctions a receive distribution budget needs.
Amplitude and Phase Balance Need End-to-End Calibration
A well-designed equal Wilkinson can provide close amplitude and phase tracking between its outputs. That is valuable when receivers compare the same antenna signal coherently. It does not make two independent receivers coherent, and it does not create antenna diversity. Shared time/frequency references, deterministic sampling, cables, filters, receiver channels and software alignment still matter.
For an array, splitting one antenna into two receivers is not the same as combining two spatial elements. When the divider is part of a calibration or reference system, characterize the entire path from the declared input plane to each digitized channel. Rohde & Schwarz’s phase-difference measurement note shows why splitter imbalance and the test setup itself must be calibrated out when precise phase comparison is required.
Measure the Three-Port Network and the Installed System
At minimum, calibrate a vector network analyser at the divider connectors and terminate every unused port in the declared impedance. Sweep:
-
S11,S22andS33for common- and output-port match; -
S21andS31for total insertion loss; - excess loss after subtracting the 3.0103 dB equal-division term;
-
S23for output-to-output isolation with the common port correctly terminated; - amplitude and phase difference between the two common-to-output paths; and
- the same quantities with representative receiver states or measured receiver impedances when mismatch sensitivity matters.
Then test the installed receive chain. Inject or receive a repeatable weak signal and calibrated noise, record SNR at each receiver with fixed bandwidth and gain, and add representative blockers. Switch receiver states, disconnect one output, and check whether the other receiver’s noise floor, spurs or gain changes. Repeat A/B/A with the alternative splitter so propagation and source drift do not become the result.
When you split, do not just divide—define the ports, preserve the noise margin, and verify the isolation.
Primary and Authoritative Technical Sources
- E. J. Wilkinson, “An N-Way Hybrid Power Divider,” 1960—the foundational equiphase/equiamplitude hybrid-divider paper.
- Mini-Circuits AN10-006: Understanding Power Splitters—division loss, isolation, mismatch and internal-resistor behaviour.
- Mini-Circuits: Resistive versus Wilkinson Splitter/Combiners—measured implementation differences and nominal 3 dB versus 6 dB paths.
- Keysight 87303C specifications—a current primary example separating division, excess insertion loss, isolation and tracking.
- Keysight: Noise Figure Measurement—available versus delivered noise power, source match and receiver contribution.
- Rohde & Schwarz: Measurement of Phase Difference—calibrating splitter/test-path imbalance in precise phase measurements.
Mini-FAQ
- When does a Wilkinson fit an RX splitter? It is attractive when low excess loss, matched ports and output isolation are needed over a defined band; a resistive, transformer, diplexer or active topology may better fit another requirement.
- Why is each output about 3 dB lower? An equal two-way divider sends half the available power to each output. The ideal division term is 3.0103 dB; conductor, dielectric, transformer and mismatch effects add excess loss.
- Does the 3 dB split automatically reduce SNR by 3 dB? No. Wanted signal and antenna noise divide together. SNR falls when excess loss and receiver-added noise become significant after the split.
- Does a Wilkinson prevent all receiver interaction? No. Isolation is finite and frequency- and load-dependent. Receiver mismatch, reverse leakage, common mode and strong-signal nonlinearity still require measurement.
- Can a transformer splitter be better at HF? Yes, when its measured bandwidth, loss, match, balance, isolation, common-mode behaviour and linearity fit the system. A transformer label alone proves none of them.
- What should be measured before installation? Measure all port matches, both insertion paths, excess loss, output isolation and amplitude/phase balance, then verify receiver SNR, spurs and blocker headroom with representative loads.