Dipole & Yagi Antenna Calculator: Length, Impedance and Pattern

Solve dipole, inverted-V, radial ground-plane and Yagi currents with a thin-wire method of moments. Find impedance, resonant length, gain, radiation patterns and VSWR bandwidth.

#antennas #dipole #ground plane #inverted-V #NEC-2 #thin-wire MoM #VSWR #Yagi

This dipole antenna calculator and Yagi antenna calculator finds the currents that make a wire antenna radiate. Choose a straight dipole, an inverted-V, a vertical ground plane with radial wires, or a Yagi with two to five elements. The tool is intended for radio amateurs, RF engineers, students, and LoRa or Meshtastic builders comparing practical antenna dimensions.

Start with the operating frequency, bare conductor diameter, and feed-line impedance. The geometry controls use fractions of the design wavelength, so changing frequency scales the antenna while retaining its proportions. The cutting schedule converts those ratios into physical lengths. For an 868 MHz antenna length, the wavelength is about 345 mm; a half-wave conductor is roughly 173 mm before resonance trimming. The default starts slightly shorter. A dipole or inverted-V length includes both arms; the ground-plane radiator length is one vertical wire.

The calculation solves a thin-wire electric-field integral equation with a piecewise sinusoidal method of moments. Each Yagi reflector and director carries an induced current determined by mutual coupling. The ground plane antenna calculator likewise solves the radiator and its equally spaced radials together, with current continuity at their junction. Wire diameter changes the kernel and therefore the input impedance and resonant length.

Read resistance and reactance together. A geometrical half-wave dipole is normally inductive: the familiar ideal reference is approximately 73 + j42.5 Ω. Resonance means zero reactance, and usually requires a shorter conductor. The resonance search changes only the driven conductor; Yagi parasitic elements, spacings and ground-plane radials retain their entered dimensions. A resonant antenna can still need impedance matching.

Use the antenna radiation pattern calculator to compare the directions of radiation. The boom points along +x and the horizontal dipole wires along y. Azimuth is measured from +x toward +y. Elevation is the x–z cut, with 90° pointing upward. Each polar is normalized separately: its outer ring is 0 dB and its centre is −30 dB or lower. Play the current phase to see the standing current reverse on the conductors; the radiation patterns are cycle averages.

The frequency sweep holds the physical antenna fixed. Its bandwidth is the contiguous interval below 2:1 VSWR around the lowest resolved VSWR minimum. Missing crossings are left blank; extend the sweep before treating that interval as a bandwidth. Yagi bandwidth is also checked on a finer current mesh and left blank if either impedance or band edges have not stabilized. Increase Spatial resolution or the frequency-solve controls when the corresponding refinement is flagged.

The optional ground is an infinite perfect conductor, included by image currents in both impedance and radiation calculations. It does not represent lossy soil, roofs, buildings or terrain. Conductors are lossless, so gain equals directivity; realized gain also includes feed mismatch. Insulation, baluns, coax radiation, matching networks, conductor losses and mounting hardware are excluded. Verify a build with an antenna analyser and measurements in its intended surroundings. Installation work belongs to a qualified person.

How it is calculated

The implementation derives the mixed-potential, weak form of the thin-wire EFIE, equivalent to Pocklington's equation after integration by parts. With the ejωte^{j\omega t} convention,

Zmn=jη0[k∬fmfn t^m ⁣⋅ ⁣t^n G ds ds′−1k∬fm′fn′G ds ds′],G=e−jkR4πR, R=∣r−r′∣2+a2.Z_{mn}=j\eta_0\left[k\iint f_m f_n\,\hat t_m\!\cdot\!\hat t_n\,G\,ds\,ds'-\frac1k\iint f'_m f'_nG\,ds\,ds'\right],\quad G=\frac{e^{-jkR}}{4\pi R},\ R=\sqrt{|\mathbf r-\mathbf r'|^2+a^2}.

The current on each segment is interpolated between its endpoint coefficients with sine functions. The dipole, inverted-V and radial models test with the same functions (Galerkin MoM). A complex LU solve gives the feed and induced currents. Exact symmetry reduces the unknown count; it does not remove mutual coupling. Opposite dipole arms share an antisymmetric outgoing-current basis; equal radial currents satisfy Kirchhoff continuity at the feed.

Two-point Gauss quadrature handles the regular interactions in these Galerkin models. The singular static self interaction is integrated analytically for the linear projection of the sine/cosine functions at the Gauss points, while the smooth dynamic remainder is integrated numerically. Refinement removes this local projection error. The equations and implementation are derived here; Balanis, 4th ed., chapter 8 provides the integral-equation framework, and NEC-2 Part III supplies independent numerical examples.

Far fields integrate the solved vector currents, including their complex phase. Accepted power is ℜ(VI∗)/2\Re(VI^*)/2; radiated power integrates intensity over equal-solid-angle strips. Peak gain is 4πUmax/Prad4\pi U_{max}/P_{rad}; realized gain includes 1−∣Γ∣21-|\Gamma|^2. The PEC image reverses horizontal current components, preserves vertical ones and reverses image charge, affecting both the impedance matrix and the upper-hemisphere field. A safeguarded secant search trims only the driven conductor to zero reactance.

The dipole, inverted-V and radial frequency scans test direct MoM midpoint solutions against linear prediction of the complex reflection coefficient. Regions with excess curvature are subdivided; every discovered local reflection minimum is refined with a bounded Brent search. The lowest resolved minimum selects the passband, whose 2:1 edges are found with bracketed true frequency solves. Shape-preserving cubic interpolation draws the frequency curves. Sparse current support is not assumed away: the exact mirror symmetry is aggregated in the reaction kernel, and the same complex system is solved through an equivalent real block system.

Geometry-dependent reaction kernels are prepared once and shared by the feed and frequency solves. Feed currents use a 1 V peak normalization; numerical matrix coefficients use SI values before the output steps restore their physical units. Radiation integration and the length search are separate from the frequency sweep, so changing phase or angular resolution does not repeat frequency solutions.

For a Yagi, the parallel-wire EFIE is integrated to its coupled Hallén form, with one integration constant per element:

∑q∫Gpq(z,z′)Iq(z′) dz′=Cpcos⁡(kz)+Vp2jη0sin⁡(k∣z∣).\sum_q\int G_{pq}(z,z')I_q(z')\,dz'=C_p\cos(kz)+\frac{V_p}{2j\eta_0}\sin(k|z|).

Only the driven element has nonzero voltage. Point matching includes every node and the zero-current tips; the current basis remains piecewise sinusoidal. Half-element nodes follow zj=(L/2)[1−(1−j/m)3]z_j=(L/2)[1-(1-j/m)^3] to resolve the end charge. All reflector/director coupling and the PEC images remain in the system. This equivalent Hallén formulation avoids the repeated derivative-reaction products of the weak form; see Orfanidis, Currents on Linear Antennas.

Collinear Yagi interactions use the circumferentially averaged static cylindrical kernel, ⟨(4πu2+4a2sin⁡2(φ/2))−1⟩φ\langle(4\pi\sqrt{u^2+4a^2\sin^2(\varphi/2)})^{-1}\rangle_\varphi. Sixteen angular Gauss points integrate its smooth remainder after extracting the logarithmic singular term. Exact source moments integrate a cubic projection of the sinusoidal basis; four-point Gauss quadrature integrates the regular dynamic remainder. That remainder and interactions with other elements retain the reduced-radius thin-wire kernel. This improves the self and near-segment integrals without imposing the segment/radius restriction on straight collinear segments.

The Yagi frequency matrix uses 8-point Chebyshev interpolation, raised to 12 if direct solves at the design frequency, best match and edges differ by more than 10⁻⁷ in complex reflection coefficient. An unresolved interpolation test withholds the band. Final threshold crossings are refined with the direct Hallén matrix. A separate finer spatial mesh checks complex impedance at the design frequency, a fixed sweep probe and the matching minimum, then directly re-solves both edges; bandwidth is reported only when the impedance change is below 3% and each edge shift is below 2% of the band width.

Assumptions

Bare, round, perfectly conducting thin wires in air; an ideal balanced delta-gap feed with 1 V peak excitation; open wire tips; no boom current, insulation, matching network or feed-line common mode. Yagi directors have equal lengths and all elements have equal spacing. Ground-plane radials have equal lengths, equal droop and uniform azimuth spacing. Geometry is symmetric, as required by the exact symmetry reduction. Current phases in the cutting schedule use the outward direction of each arm, so opposite dipole arms differ by 180°. The optional ground is infinite PEC, not soil.

Where it applies

Use for electrically thin, isolated conductors near their fundamental resonance. Diameter must remain below 0.02 wavelength at the highest swept frequency. In the reduced-kernel dipole, inverted-V and radial models, each segment must exceed 2.5 radii; about 8 radii is preferable (NEC-2 Part III, pp. 7–9). Yagi collinear static interactions use the cylindrical singular kernel instead, allowing short end segments. Refine the mesh and compare results before cutting a final antenna. The finite-radius MoM is not forced to equal the ideal sinusoidal 73 + j42.5 Ω approximation; feed idealization and discretization produce a few percent difference in the reference comparisons.

The default dipole uses 16 segments per arm, an initial 61-point frequency scan, 3,200 solid-angle directions and 361 points per polar. Ground-plane cases scale the per-arm mesh with radial count, with an additional factor of 0.75 above PEC. Yagis use cubic end grading and 0.75 times the Spatial resolution control, rounded upward, with a minimum of 6 segments per arm; this count is independent of element count and ground selection. Their check mesh adds the larger of 4 segments or half the main count. Four- and five-element Yagis reduce the solid-angle grid by √(3 / element count) to reserve computation for resolving narrow frequency features. The actual main/check spatial, angular and adaptive frequency counts are listed in How it's calculated. Spatial resolution is limited to 24 to keep the largest refinement runs inside the browser limit; multi-element checks can still take several seconds. A finite grid cannot certify all geometries. No mutual interactions or field components are removed.

The adaptive scan adds current-system frequency solutions at midpoints, near all discovered matching minima, and at bracketed threshold crossings. The complex-reflection curvature target is 0.08√(13 / initial count); the crossing frequency-ratio tolerance is 2×10⁻⁶ × 13 / initial count. The minimum search uses ten times that frequency tolerance. Search depth is limited to nine subdivisions and discovery to 6 × initial count + 80 solves. If discovery, minimum refinement or a crossing remains unresolved, the bandwidth and uncertain edges are deliberately blank: increase Initial sweep samples or narrow the sweep span. A finite scan cannot prove absence of arbitrarily narrow unsampled features; spatial and frequency refinement remain necessary. Yagi matrix interpolation is independently checked at the selected band and its reported crossings use direct matrices.

The plotted frequency curves use shape-preserving cubic interpolation through the actual, nonuniform solved frequencies, supplemented by 241 evenly spaced drawing points. Reported bandwidth edges come from direct solves, not from the drawn interpolation. Spatially unresolved Yagi bandwidth and both edges are deliberately blank, even when a frequency crossing is found. Its main and check meshes must agree within 3% in complex impedance and within 2% of the current band width at each edge; these are numerical acceptance tolerances, not measurement accuracy. Increase Spatial resolution when flagged. Near a merging or disappearing passband, no bandwidth is reported.

The resonance search covers 0.34–0.58 wavelength total driven length, or 0.15–0.31 for the monopole, and holds everything else fixed. Its bracket is further limited by wire-to-ground clearance and, for the reduced-kernel models, the segment-to-radius condition. An unresolved resonance or sweep edge is deliberately blank.

Polars show relative power down to −30 dB, normalized separately; gain uses the full spherical or upper-hemisphere integration. In free space the azimuth cut is horizontal; above PEC it uses the strongest elevation in the +x vertical cut. The elevation cut may differ from a Yagi E-plane when the wire orientation changes. The wire thickness and transverse current-envelope displacement are enlarged for visibility; colours and the envelope come from the solved current. A phase cycle is shown in slow motion. Finite conductivity, soil absorption, RF exposure, transmitter limits, structural loading and lightning protection are outside this model.

Wire positions use a projected scale and origin set by the allowed wire-length and angle ranges, the selected boom geometry and feed height. This coordinate frame stays fixed while the length or angle handle moves, independently of the solved current envelope. Current-wave displacement is exaggerated and limited uniformly to the reserved envelope space; closely spaced Yagis reduce it further to keep adjacent envelopes apart. Colour and relative envelope heights still come from the solved complex currents.

References

  • Balanis, C. A., Antenna Theory: Analysis and Design, 4th edition (2016), chapter 8. ISBN 9781118642061. https://www.wiley.com/en-us/Antenna%2BTheory%3A%2BAnalysis%2Band%2BDesign%2C%2B4th%2BEdition-p-9781118642061
  • Burke and Poggio, NEC-2, Part III: User’s Guide, LLNL; wire modeling and Examples 1–2, printed pp. 7–9 and 83–90. https://www.nec2.org/other/nec2prt3.pdf
  • Viezbicke, P. P., Yagi Antenna Design, NBS Technical Note 688 (1976), Table 1 and pattern comparisons. https://doi.org/10.6028/NBS.TN.688
  • IEEE EMC TC-9, Dipole Antenna Input Impedance Canonical Validation Problem 5.2.1.1. The text calls the half-wave case “resonant” although its quoted +j42.5 Ω is not zero reactance; here half-wave length and resonance are distinguished. https://ewh.ieee.org/cmte/tc9/Problems/problem5-2-1-1-DipoleInputImpedance-PROBLEM-DESCRIPTION.pdf
  • Orfanidis, S. J., Electromagnetic Waves and Antennas, open author textbook, chapter 24, Currents on Linear Antennas: coupled Hallén equations, piecewise sinusoidal basis and end conditions. https://eceweb1.rutgers.edu/~orfanidi/ewa/ewa-1up.pdf

Inputs

InputSymbolUnitDefault
AntennaantennaDipole
Frequencyf0MHz868
Wire diameterd_wiremm1.5
Feed-line impedanceZ_feedohm50
Driven length / wavelength
Total tip-to-tip conductor length, including both arms.
L_ratio0.48
Radiator length / wavelengthH_ratio0.24
Included V angleV_angledeg120
Radial length / wavelengthR_ratio0.24
RadialsN_rad4
Radial droop
Angle below the horizontal plane.
droopdeg30
Yagi elements
One reflector, one driven element and up to three directors.
N_el3
Reflector length / wavelengthL_ref0.5
Director length / wavelengthL_dir0.45
Element spacing / wavelength
Equal spacing along the +x boom direction.
spacing0.2
EnvironmentenvironmentFree space
Feed height / wavelength
Height of the feed or V apex; radials and wire tips must clear the ground.
height_ratio0.5
Current phasephase_currentdeg25
Spatial resolution
Refine this control; the actual main and check segment counts are listed in the calculation.
mesh16
Initial sweep samples
Sets scan density and refinement tolerance; extra solves resolve minima and band edges.
sweep_count61
Sweep half-span / frequencysweep_span0.12
Elevation integration strips
Radiated power uses twice as many azimuth samples.
angular_grid40

Results

With the default inputs above, the calc gives:

ResultValue
Driven conductor length165.784 mm
Resonant driven length162.79 mm
Input resistance77.191 Ω
Input reactance13.249 Ω
Feed-line VSWR1.62
Peak gain (dBi)2.152 dB
Realized gain (dBi)1.902 dB
Front-to-back, +x / −x0 dB
2:1 VSWR bandwidth79.399 MHz
2:1 lower edge808.787 MHz
2:1 upper edge888.186 MHz
Azimuth-cut elevation0 deg
Bandwidth coverageBoth edges resolved
Resonance searchResonance found
Best-match frequency845.624 MHz
Lowest swept VSWR1.424

Checked against sources

Each case below was calculated by this calc and agrees with the value from its source (within the tolerance where one is shown).

Ideal thin half-wave dipole benchmark

Source: IEEE EMC TC-9 canonical problem 5.2.1.1, pp. 1–2, ideal sinusoidal current and vanishing-radius approximation. The finite-radius Galerkin solution is compared with 4% resistance and 2.5% reactance tolerance, not calibrated to the ideal current. The source's word 'resonant' is inconsistent with +j42.5 Ω; this is the geometric half-wave case.

Inputs: Frequency = 1 MHz; Wire diameter = 0.001 mm; Driven length / wavelength = 0.5; Initial sweep samples = 13; Elevation integration strips = 24; the others as above.

ResultSourceThis calc
Input resistance73 ± 2.92 Ω75.326 Ω
Input reactance42.5 ± 1.063 Ω42.318 Ω

NEC-2 finite-radius half-wave example

Source: NEC-2 Part III Example 1, pp. 83–84: length 0.5 m, radius 0.001 m, wavelength 1 m, 7 NEC segments. The 6% tolerance allows the different delta-gap versus finite-segment feed and reduced-kernel self quadrature. Expected values are the independent NEC listing.

Inputs: Frequency = 299.8 MHz; Wire diameter = 2 mm; Driven length / wavelength = 0.5; Initial sweep samples = 13; Elevation integration strips = 24; the others as above.

ResultSourceThis calc
Input resistance82.698 ± 4.962 Ω85.458 Ω
Input reactance46.306 ± 2.778 Ω44.379 Ω

NEC-2 electrically short dipole

Source: NEC-2 Part III Example 2, pp. 87–90: length 0.5 m, radius 10 μm, 200 MHz, 8 segments and current-slope-discontinuity source. Differences in feed treatment and quadrature warrant a 6% comparison tolerance; reference values are not generated by this solver.

Inputs: Frequency = 200 MHz; Wire diameter = 0.02 mm; Driven length / wavelength = 0.3336; Initial sweep samples = 13; Elevation integration strips = 24; the others as above.

ResultSourceThis calc
Input resistance26.576 ± 1.595 Ω25.637 Ω
Input reactance-632.1 ± 37.924 Ω-600.6 Ω

NBS three-element Yagi gain scale

Source: Viezbicke NBS TN 688 (1976), Table 1, printed p. 7 (PDF p. 17): 0.4λ Yagi, 0.482λ reflector, 0.424λ director, diameter 0.0085λ, measured 7.1 dB relative to a half-wave dipole at the same height. Adding 2.15 dB gives a 9.25 dBi scale benchmark. The driven length is unspecified and the reference involves a ground-height comparison, so the 8% tolerance in dB (0.74 dB) checks gain scale, not replication of that measured installation.

Inputs: Antenna = Yagi; Frequency = 400 MHz; Wire diameter = 6.371 mm; Driven length / wavelength = 0.46; Reflector length / wavelength = 0.482; Director length / wavelength = 0.424; Initial sweep samples = 13; Elevation integration strips = 24; the others as above.

ResultSourceThis calc
Peak gain (dBi)9.25 ± 0.74 dB8.948 dB

Related Resources

Responses

Your email address will not be published. Required fields are marked *

L ading...