Cylindrical Wave Expansion
Modal Structure of the Cylindrical Field
The expansion rests on separation of variables applied to the scalar Helmholtz equation in circular cylindrical coordinates. Each field component separates into a product of three factors: a radial function in ρ, an azimuthal function in φ, and an axial function in z. The azimuthal factor must be single valued as φ increases by 2π, which forces it into the integer harmonic series ejnφ with n = 0, ±1, ±2 and so on. The axial factor e−jkzz is not quantized because the cylinder is open along its axis, so kz sweeps a continuum from −k to +k for propagating contributions and beyond ±k for evanescent ones. The remaining radial separation constant is fixed by kρ2 = k2 − kz2.
For a transmitting antenna the exterior radial dependence must represent energy flowing outward and decaying, so the Hankel function of the second kind Hn(2)(kρρ) is the correct choice under the ejωt convention. Its large-argument asymptotic form falls off as 1/√ρ, the signature of a two-dimensional cylindrical wave, in contrast to the 1/r decay of a spherical wave. The complete vector field is built from two independent polarizations, the transverse-electric and transverse-magnetic cylindrical modes, each weighted by complex coefficients bn(kz) that encode everything measurable about the source.
Practical use almost always pairs the expansion with sampled near-field data. A probe traces a measurement cylinder of radius ρ0 while recording amplitude and phase on a grid in φ and z. A fast Fourier transform in φ recovers the azimuthal coefficients and a second FFT in z recovers the axial spectrum, after which probe correction and analytic continuation through the Hankel functions propagate the field to any larger radius, including the far zone.
Governing Equations
E(ρ,φ,z) = ∑n=−∞∞ ∫−∞∞ bn(kz) Hn(2)(kρρ) ejnφ e−jkzz dkz
Radial wavenumber constraint:
kρ2 = k2 − kz2, k = 2π/λ
Outgoing asymptotic form (large kρρ):
Hn(2)(kρρ) ≈ √(2 / πkρρ) × e−j(kρρ − nπ/2 − π/4)
Azimuthal mode truncation rule:
N = ⌈kρmax⌉ + 10, n ∈ [−N, +N]
Where kz is the axial wavenumber, kρ the radial wavenumber, ρmax the largest transverse radius enclosing the currents, and bn(kz) the modal coefficients. Example: an aperture of 0.5 m diameter at 10 GHz (λ = 30 mm, k ≈ 209 rad/m) has ρmax = 0.25 m, so kρmax ≈ 52, giving N ≈ 62 and 2N+1 ≈ 125 azimuthal modes.
Comparison of Modal Expansions
| Expansion | Coordinate basis | Radial / range basis | Decay | Angular coverage | Best suited antenna |
|---|---|---|---|---|---|
| Cylindrical | (ρ, φ, z) | Hankel Hn(2) | 1/√ρ | Full 360° azimuth, limited elevation | Base station panels, fan-beam radar |
| Spherical | (r, θ, φ) | Spherical Hankel hn(2) | 1/r | Complete 4π sphere | Broad-beam, low-gain, general |
| Planar | (x, y, z) | Plane-wave spectrum | None (FFT) | Forward cone only | High-gain pencil-beam arrays |
| Cylindrical (evanescent) | (ρ, φ, z) | Hn(2), |kz| > k | Exponential | Reactive near zone | Coupling and probe correction |
Frequently Asked Questions
How many cylindrical modes are needed to represent an antenna of a given size?
The azimuthal index runs from −N to +N with N ≈ ⌈kρmax⌉ + 10, where ρmax is the largest transverse radius enclosing the currents. An aperture of 0.5 m diameter at 10 GHz has ρmax = 0.25 m and kρmax ≈ 52, so N ≈ 62 and about 125 (2N+1) azimuthal modes. The axial spectrum is sampled over ±k with Δz ≤ λ/2 by Nyquist. Too few modes alias the far field; too many only add noise-amplifying evanescent terms.
Why are Hankel functions used for the radial dependence instead of Bessel functions?
Under the ejωt convention, Hn(2)(kρρ) asymptotically behaves as an outward-traveling wave decaying as 1/√ρ, satisfying the radiation condition for an exterior transmitting field. A plain Bessel function Jn is a standing wave valid only in a source-free interior. Using the wrong kind, or the Hankel function of the first kind with this time convention, gives a field that grows with distance and is unphysical.
How does cylindrical wave expansion compare to spherical and planar expansions for near-field measurement?
A cylindrical scan captures full 360° azimuth and suits long or fan-beam antennas like base station panels, using Hankel functions radially, a Fourier series in φ, and an FFT in z. Planar expansion is fastest for high-gain pencil beams but truncates outside a cone. Spherical expansion covers the complete 4π sphere at higher cost. Cylindrical scanning is the practical middle ground for full azimuth with limited elevation.