Skip to content
FiftyOhm
Esc
  • Air-core coilSingle-layer air-core solenoid inductance from Nagaoka's exact coefficient, compared with Wheeler's formula. Solve for the number of turns for a target inductance.Components & magnetics
  • Attenuator PadsResistor values for Pi, T, bridged-T and minimum-loss L-pad attenuators, with unequal impedances, power dissipation per resistor and E24/E96 rounding.Couplers, dividers & attenuators
  • CoaxCoax Z0, capacitance and inductance per length, velocity factor, loss, exact TE11 cutoff and breakdown voltage. Solve for D, d or εr from Z0.Transmission lines
  • CPW / GCPWCoplanar waveguide and grounded CPW impedance, effective permittivity, dispersion, delay and loss by conformal mapping. Solve for centre width or gap.Transmission lines
  • dBm ↔ W ↔ VConvert RF power between dBm, dBW, watts, Vrms, Vpeak, Vpp, dBµV and dBmV for any system impedance. Instant, accurate, with formulas.Fundamentals & conversions
  • Differential pairEdge-coupled microstrip and stripline differential impedance, Z0 even/odd, common-mode impedance, coupling and delay. Solve for spacing or width from Zdiff.Transmission lines
  • Filter OrderMinimum filter order to meet a passband/stopband attenuation mask for Butterworth, Chebyshev and elliptic responses, with the attenuation actually achieved.Filters
  • Free-space path lossFree-space path loss from the Friis equation, received power with antenna gains, and a far-field check. Solve for distance or frequency.Propagation & link budget
  • L / Pi / T MatchingDesign lumped L, Pi and T matching networks between complex source and load impedances. Both L solutions, loaded-Q Pi/T, component values and bandwidth.Matching & Smith chart
  • LC Filter DesignerDesign lumped LC ladder filters: Butterworth, Chebyshev or Bessel; low-pass, high-pass, band-pass, band-stop to order 10, with exact insertion loss and group delay.Filters
  • MicrostripMicrostrip Z0, effective permittivity, dispersion, delay and loss. Solve for trace width from target impedance. Hammerstad–Jensen + Kirschning–Jansen models.Transmission lines

Wilkinson Power Divider Calculator

Equal and unequal-split Wilkinson divider: arm impedances, isolation resistor, output transformers, λ/4 lengths and lumped L-C equivalents.

3 cited sources19 reference checksRuns in your browser

Inputs

1 = equal split (−3.01 dB each). 0.5 sends twice as much power to port 2 as to port 3.

1 for air lines. For microstrip, take εeff for each line width from the Microstrip Impedance calculator.

Results

Arm impedance to port 2(Z₀₂)
70.711
Arm impedance to port 3(Z₀₃)
70.711
Isolation resistor(R)
100
Quarter-wave physical length(λg/4)
19.005
Power to port 2 (relative to input)(P2/P1)
−3.0103dB
Power to port 3 (relative to input)(P3/P1)
−3.0103dB
Lumped arm 2: series inductor(L₂)
4.6891
Lumped arm 2: shunt capacitor at each end(C₂)
937.83
Lumped arm 3: series inductor(L₃)
4.6891
Lumped arm 3: shunt capacitor at each end(C₃)
937.83
Lumped: combined shunt capacitor at port 1(C₂+C₃)
1.8757
  • A single-section Wilkinson is perfectly matched and isolated only at f₀; both degrade away from f₀ (the lumped version faster). Use multi-section designs for octave bandwidths.

Formulas

K2=P3P2,Z03=Z01+K2K3,Z02=K2Z03=Z0K (1+K2)K^2 = \frac{P_3}{P_2},\qquad Z_{03} = Z_0\sqrt{\frac{1+K^2}{K^3}},\qquad Z_{02} = K^2 Z_{03} = Z_0\sqrt{K\,(1+K^2)}R=Z0(K+1K),R2=Z0K,R3=Z0KR = Z_0\left(K+\frac{1}{K}\right),\qquad R_2 = Z_0K,\quad R_3 = \frac{Z_0}{K}Z04=R2Z0,Z05=R3Z0Z_{04} = \sqrt{R_2 Z_0},\qquad Z_{05} = \sqrt{R_3 Z_0}ℓ=λg4=c04f0εeff\ell = \frac{\lambda_g}{4} = \frac{c_0}{4 f_0\sqrt{\varepsilon_\mathrm{eff}}}λ/4 line Z → Π equivalent at f0:ω0L=Z,ω0C=1Z\lambda/4\text{ line }Z\ \to\ \Pi\text{ equivalent at }f_0:\quad \omega_0 L = Z,\qquad \omega_0 C = \frac{1}{Z}

What this calculator does

A Wilkinson power divider splits the power entering port 1 between ports 2 and 3. All three ports are matched, and ports 2 and 3 are isolated from each other. It is built from two quarter-wave (λ/4\lambda/4) transmission-line arms and one resistor RR between the ends of the arms. When the outputs are driven in phase (the even mode), both ends of RR sit at the same voltage, so no current flows in it and the divider is lossless. A signal that enters port 2 or port 3 (the odd mode) puts a virtual ground halfway along RR, and the resistor absorbs the power. That is why a reflection from one output does not reach the other output.

For a power ratio K2=P3/P2K^2 = P_3/P_2, the calculator gives you:

  • the arm impedances Z02Z_{02} and Z03Z_{03} and the isolation resistor RR (Pozar §7.3);
  • for an unequal split, the impedances R2=Z0KR_2 = Z_0K and R3=Z0/KR_3 = Z_0/K that the arms see, plus the output quarter-wave transformers Z04=R2Z0Z_{04} = \sqrt{R_2Z_0} and Z05=R3Z0Z_{05} = \sqrt{R_3Z_0} that bring both outputs back to Z0Z_0;
  • the physical quarter-wave length λg/4=c0/(4f0εeff)\lambda_g/4 = c_0/(4f_0\sqrt{\varepsilon_\text{eff}});
  • a lumped-element equivalent for each λ/4\lambda/4 line, for low frequencies where a quarter-wave line would be too long.

For an equal split (K=1K = 1), the equations reduce to the familiar Z02=Z03=2 Z0=70.71 ΩZ_{02} = Z_{03} = \sqrt{2}\,Z_0 = 70.71\ \Omega and R=2Z0=100 ΩR = 2Z_0 = 100\ \Omega in a 50 Ω system. No output transformers are needed.

Effective permittivity

The quarter-wave length depends on the effective permittivity of the line, not on the substrate's εr\varepsilon_r. Get εeff\varepsilon_\text{eff} from the Microstrip Impedance calculator: solve the trace width for each arm impedance, then read off εeff(f)\varepsilon_\text{eff}(f). Arms with different impedances have different widths, so their εeff\varepsilon_\text{eff} differ too. A 70.7 Ω line has a slightly lower εeff\varepsilon_\text{eff} than a 50 Ω line on the same board. For an unequal split, run the length calculation once for each line with its own εeff\varepsilon_\text{eff}. Use 1 for air lines. For stripline, use εr\varepsilon_r.

Lumped-element equivalent

At the design frequency, a λ/4\lambda/4 line of impedance ZZ has the ABCD matrix [0jZj/Z0]\begin{bmatrix}0 & jZ\\ j/Z & 0\end{bmatrix}. A Pi network made of a series inductor and two shunt capacitors has exactly the same matrix when

ω0L=Z,ω0C=1Z.\omega_0 L = Z, \qquad \omega_0 C = \frac{1}{Z}.

The equivalence holds exactly at f0f_0 and only approximately away from it. In a lumped Wilkinson, the two shunt capacitors at port 1 are in parallel, so the calculator also gives their sum C2+C3C_2 + C_3. A T-network equivalent (series L, shunt C, series L) has the same element values and is sometimes easier to lay out.

Solving for the split ratio

Click ⇄ Solve next to Power split ratio P3/P2 and choose a target. This answers questions such as "which split gives a 100 Ω arm?" (target Z03Z_{03}) or "what ratio puts −4.77 dB at port 3?" (target the port-3 power). Z02Z_{02} rises and Z03Z_{03} falls monotonically with KK, so the solution is unique. You can also solve the frequency for a target length or inductor value. That is useful when a layout fixes the line length.

Worked example

Take an equal split in a 50 Ω system at 2.4 GHz, with εeff=2.7\varepsilon_\text{eff} = 2.7 (the defaults). The calculator gives:

QuantityValue
Z02=Z03Z_{02} = Z_{03}70.711 Ω
RR100 Ω
Power to each output−3.0103 dB
λg/4\lambda_g/419.005 mm
Lumped series LL per arm4.6891 nH
Lumped shunt CC at each end of an arm0.93783 pF
Combined CC at port 11.8757 pF

Now change the ratio to 0.5, so port 2 gets twice the power of port 3. The arms become Z02=51.494Z_{02} = 51.494 Ω and Z03=102.99Z_{03} = 102.99 Ω, and the isolation resistor becomes R=106.07R = 106.07 Ω. The arms end in R2=35.355R_2 = 35.355 Ω and R3=70.711R_3 = 70.711 Ω, so the output transformers are Z04=42.045Z_{04} = 42.045 Ω and Z05=59.460Z_{05} = 59.460 Ω. The outputs receive −1.7609 dB and −4.7712 dB.

Accuracy and validity

The design equations are exact for ideal, lossless TEM lines at the design frequency. What limits a real divider is everything the equations leave out:

EffectTypical impactWhat to do
BandwidthA single section is perfect only at f0f_0; match and isolation degrade either side (Pozar plots the response in §7.3)Use two or more sections (Cohn) for an octave or more
T-junction and bend discontinuitiesShift the centre frequency by a few percentSimulate the junction in an EM solver and trim the line lengths
Resistor parasiticsReduce isolation at high GHzUse 0402 or smaller thin-film parts and bring the arm ends close together
Very high or very low arm impedancesLines narrower than about 0.1 mm or wider than about λ/8The calculator warns outside about 20–120 Ω
Lumped versionNarrower bandwidth than the distributed one; component Q adds lossUse high-Q air-core inductors and C0G capacitors

The calculator flags arm or transformer impedances above 120 Ω or below 20 Ω. In microstrip, these are hard to etch or behave poorly at the junction.

Practical notes

  • Port 2 and port 3 are not interchangeable in an unequal design. The arm with the higher impedance goes to the port that receives less power.
  • Isolation needs the resistor to be physically short. The two arm ends must meet at the resistor pads. A long route to the resistor adds phase in the odd mode and degrades isolation.
  • Power handling. In normal operation, RR dissipates nothing. If one output is open or shorted, up to half the input power can end up in RR, so size it for the fault case.
  • Unequal splits beyond about 4:1 need arm impedances above about 130 Ω. Consider a coupled-line or lumped design, or stack a Wilkinson with an attenuator instead.
  • Combiner use. The same circuit combines coherent, equal-phase signals without loss. Incoherent signals lose half their power in RR.

References

  1. D. M. Pozar, Microwave Engineering, 4th ed., Wiley, 2012, §7.3 “The Wilkinson Power Divider,” Eqs. (7.37)–(7.38).
  2. E. J. Wilkinson, “An N-Way Hybrid Power Divider,” IRE Trans. Microwave Theory Tech., vol. MTT-8, no. 1, pp. 116–118, Jan. 1960.
  3. L. I. Parad and R. L. Moynihan, “Split-Tee Power Divider,” IEEE Trans. Microwave Theory Tech., vol. MTT-13, no. 1, pp. 91–95, Jan. 1965 (unequal split).

This calculator's results are checked against 19 independent reference values in automated tests on every change. How we check accuracy.