Antenna Technology

Corporate Feed Network

Q: How does a corporate feed differ from a series feed?

A corporate feed splits the input through a symmetric binary tree of power dividers so that every element sees the same physical path length, which makes the inter-element phase nearly independent of frequency and produces a broadside beam that does not squint as frequency shifts. A series feed taps elements off a single traveling-wave line, so the progressive phase between elements scales with electrical line length and the beam scans with frequency. The corporate feed therefore supports much wider instantaneous bandwidth, typically 15 to 40 percent versus a few percent for a series feed, at the expense of more line length, higher distribution loss, and larger routing area. Corporate feeds are preferred for fixed broadside arrays, electronically steered phased arrays where each port carries an independent phase shifter, and any application demanding stable patterns over a wide band.

Q: Why does corporate feed loss increase with array size?

In an N-element corporate feed the signal traverses log2(N) divider stages, and each stage adds both the divider's split loss and the loss of the connecting transmission line. While an ideal lossless 2-way split is 3 dB by design (the power is divided, not dissipated), real microstrip or stripline at millimeter-wave frequencies adds dissipative loss of roughly 0.1 to 0.5 dB per centimeter, and the total guided path length grows with array span. A 64-element feed at 30 GHz can accumulate 3 to 6 dB of one-way insertion loss in the network alone, directly degrading transmit EIRP and receive noise figure. Designers mitigate this with low-loss substrates such as PTFE composites or fused silica, air-filled suspended stripline, or by partitioning the array into subarrays so amplification occurs before the lossy distribution.

Q: How is amplitude tapering implemented in a corporate feed?

A uniform corporate feed uses equal-split dividers so every element receives the same power, which yields the narrowest beam but high -13.3 dB sidelobes. To lower sidelobes, designers replace the equal-split dividers with unequal-ratio dividers whose coupling ratios follow a window function such as Taylor, Chebyshev, or binomial. For example, a Taylor n-bar distribution might feed the array center with several dB more power than the edges, suppressing the first sidelobe to -25 or -30 dB at the cost of a slightly wider main beam. Because the power ratios are fixed by the divider geometry, tapered corporate feeds give a static illumination; reconfigurable tapering instead requires variable attenuators or gain control at each element, which is common in active electronically scanned arrays.

/KOR-puh-rit feed NET-wurk/

Built as a symmetric binary tree of power dividers, this parallel array-feed architecture splits one input into 2, 4, 8, or more branches so that every radiating element sits at the end of an electrically identical path. Because the guided length to each element is equal, the relative phase across the aperture stays nearly constant with frequency, giving a stable broadside beam and wide instantaneous bandwidth. Corporate feeds dominate fixed-beam panel antennas and electronically steered phased arrays, where each output port drives an independent phase shifter. The trade-off versus a series feed is greater total line length, higher distribution loss, and more routing area, which becomes the limiting factor at millimeter-wave frequencies.

Category: Antenna Technology

Topology: Binary divider tree

Bandwidth: 15 to 40%

How Equal-Path Feeding Stabilizes the Beam

The defining property of a corporate feed is path symmetry. Starting from the array input port, the signal passes through a cascade of two-way power dividers, doubling the number of branches at each stage until there is one branch per radiating element. If the layout is mirror-symmetric, the physical and electrical lengths from the input to every element are identical. That equality means the inter-element phase is set only by the spatial position of the elements and any deliberate phase shifters, not by the feed line itself, so the array radiates a true broadside beam that does not squint as the operating frequency changes. This is the central advantage over a traveling-wave series feed, where the progressive phase scales directly with electrical line length and the beam scans across the band.

A feed for N elements requires log₂(N) divider stages, so a 16-element linear array uses 4 stages and a 256-element planar array uses 8. The number of two-way splitters in a fully populated binary tree is N minus 1. Each split is commonly realized with a microstrip T-junction or, where port-to-port isolation and matched terminations matter, a Wilkinson divider. The Wilkinson form prevents reflections from a failed or mismatched element from corrupting the rest of the aperture, which is why active arrays almost always pay the area cost of isolation resistors at every node.

The price of this regularity is loss and real estate. Total guided length grows with both the number of stages and the physical span of the aperture, and every centimeter of line dissipates power. At Ka-band and above, the network can swallow several dB before the signal ever reaches an element, which is why RF Essentials integrates low-noise amplifiers and drivers at the subarray level rather than feeding a large passive tree from a single source.

Governing Relationships

Number of divider stages:
S = log₂(N) → N = 2^S

Two-way splitters in a full tree:
N_div = N − 1

Ideal split ratio and isolation:
P_out = P_in / 2 (−3 dB per equal split); Wilkinson isolation R = 2Z₀ ≈ 100 Ω for 50 Ω ports

One-way distribution loss to an element:
L_net ≈ α × 𝓁_path (dB), where α ≈ 0.1 to 0.5 dB/cm at mmWave

N = element count, S = stages, Z₀ = line impedance, α = line attenuation, 𝓁_path = one-way guided length from input to an element (summed over the S segments that signal traverses). The 3 dB per split is power division, not dissipation; L_net captures the real ohmic and dielectric loss along that path.

Corporate Feed vs. Series Feed

Attribute	Corporate Feed	Series Feed	Subarray (Hybrid)
Path lengths	Equal to all elements	Progressive (cumulative)	Equal within subarray
Beam vs. frequency	Stable, no squint	Scans with frequency	Mostly stable
Instantaneous BW	15 to 40%	2 to 5%	10 to 25%
Distribution loss	Higher (long tree)	Lower (single line)	Moderate
Routing area	Large	Compact	Moderate
Best fit	Phased arrays, wideband panels	Narrowband fixed arrays	Large mmWave AESAs

Common Questions

Frequently Asked Questions

How does a corporate feed differ from a series feed?

A corporate feed splits the input through a symmetric divider tree so every element has an equal path length, keeping inter-element phase nearly frequency-independent and producing a broadside beam that does not squint. A series feed taps elements off one traveling-wave line, so phase scales with line length and the beam scans with frequency. The corporate feed supports 15 to 40% instantaneous bandwidth versus a few percent for a series feed, trading more line length, higher loss, and larger area for that stability.

Why does corporate feed loss increase with array size?

The signal crosses log₂(N) divider stages, and each stage adds line loss on top of the power split. The ideal −3 dB per equal split is division, not dissipation, but real microstrip or stripline adds roughly 0.1 to 0.5 dB/cm at mmWave, and total path length grows with aperture span. A 64-element feed at 30 GHz can accumulate 3 to 6 dB of one-way loss, hurting EIRP and noise figure. Designers fight this with low-loss substrates, suspended stripline, or subarray amplification.

How is amplitude tapering implemented in a corporate feed?

A uniform feed uses equal-split dividers, giving the narrowest beam but high −13.3 dB sidelobes. To lower sidelobes, the equal-split nodes are replaced with unequal-ratio dividers whose coupling follows a Taylor, Chebyshev, or binomial window, feeding the center more power than the edges to reach −25 to −30 dB sidelobes. Because the ratios are fixed in metal, the taper is static; reconfigurable tapering instead needs variable attenuators or per-element gain control, as in active arrays.

Related Terms

← Corporate Feed Corrected Data →