Antenna Farms & Multi-Antenna Systems

Stacking, vertical / horizontal arrays, switching matrices, automatic antenna selectors, phased verticals, beverage arrays, contest-grade installations — the multi-antenna posture

Section	Topic
1	About this volume
2	The multi-antenna posture — why stack or array
3	Vertical stacking — 2-bay, 3-bay, 4-bay Yagis
4	Horizontal stacking — phasing for fixed-direction gain
5	Phased verticals — broadside, end-fire, cardioid
6	Beverage arrays for low-band receive
7	Switching matrices — Top Ten Devices, Array Solutions
8	Automatic antenna selectors per band
9	Coupling networks — Christman, Lewallen, hybrid
10	SO2R and multi-Op station design
11	DIY build — a 2-element broadside-phased vertical array
12	Commercial buys
13	Common gotchas and myths
14	Resources

1. About this volume

Where Vols 6-15 walked through individual antennas one type at a time, and Vol 30 (Multi-radio shared) covered the inverse problem of sharing a single antenna across multiple radios via diplexers and band-switching matrices, this volume covers the third axis: using multiple antennas in service of one radio’s gain, pattern, or direction-control needs. This is the antenna-farm posture — the step that separates a respectable tower-with-one-Yagi station from a competitive contest entry, a credible EME signal on 432 MHz, a low-band DX operation that hears what no single beverage could, or a direction-finding receiver that resolves an emitter to within a couple of degrees.

Four distinct things travel under the same “multi-antenna” umbrella, and they are not interchangeable. Stacking is the use of two or more identical antennas in geometric formation to multiply aperture — two Yagis spaced correctly add roughly 2.5 dB over a single, four add roughly 5 dB, and the cost is mostly mechanical: tower, hardware, harness. Phased arrays for TX use multiple radiating elements (typically quarter-wave verticals on low bands where towers and Yagis are impractical) fed with deliberate phase relationships to synthesize directional patterns from omnidirectional radiators — the 4-square is the canonical example, the 8-circle is its scaled-up sibling. Phased arrays for RX are receive-only constructions (beverage arrays, K9AY 4-squares, Hi-Z circles, 9-circles) optimized for noise rejection and pattern steerability rather than gain; they require dedicated receive feedlines and a transceiver with a separate receive port. Switching matrices are the connective tissue that lets an operator (or a computer) select among multiple antennas mid-QSO or mid-contest, and they range from a $150 manual six-position knob to a $3000 microprocessor-controlled relay farm wired into a logging program’s band-decoder.

The hard prerequisites are real estate (a 4-square on 80 m occupies a 65-foot square; a 4-stack Yagi on 144 MHz wants a 100-foot tower for its bottom bay to be in the clear), money (commercial 4-square controllers are $800-2000; serious switching matrices are $1200-3000; a contest-grade stacking harness with vacuum relays is $500-1500 in parts alone), and willingness to spend a season learning the interactions (phasing errors create nulls instead of gain; relay-matrix insertion loss accumulates; ground systems for low-band arrays dwarf the antenna budget). For tjscientist holding Amateur Extra-class authorization and operating on his licensed bands up to 1.5 kW PEP, the upside is direct: 3-6 dB of stacking gain is worth one full S-unit on the other end, an 8-circle is the difference between answering a CW contest exchange and missing it, and a properly phased beverage array on 160 m hears Pacific signals through a Northeast-US summer thunderstorm-noise floor that an inverted-L can’t penetrate.

The four sister readings to keep open as you work through this volume are Vol 11 (Yagi-Uda) for the single-Yagi reference against which stacking gain is measured, Vol 8 (Fixed vertical monopoles) for the radiator element of every phased-vertical array, Vol 15 (Receive-only loops) for the single-beverage and K9AY baselines, and Vol 28 (Antenna modeling) for predicting array patterns before committing concrete and aluminum.

2. The multi-antenna posture — why stack or array

The dB math behind stacking is straightforward but routinely misunderstood. Doubling the aperture of an antenna system (which is what adding a second identical antenna at the right spacing accomplishes) doubles the power captured (or radiated) in the favored direction. That is +3 dB in free space, no more, regardless of how many antennas. The geometric series is therefore +3 dB for 2 antennas, +6 dB for 4, +9 dB for 8, +12 dB for 16 — each doubling adds one more 3 dB step. Real-world measurements come in 0.2 to 0.5 dB below the theoretical number due to mutual-coupling effects (the antennas interact with each other), harness loss (a 1 dB-loss phasing harness eats one full dB of the stacking gain), and the imperfect ground reflection that turns the idealized free-space pattern into a real-world elevation lobe that may or may not match the propagation angle to the target. A practical 2-stack of 5-element 144 MHz Yagis at half-wavelength spacing typically delivers +2.5 dB over the single antenna in a controlled measurement; a 4-stack delivers +5 dB; an 8-stack delivers +7.5 dB. The 0.5 dB per doubling that gets eaten by mutual coupling, harness loss, and ground effects is the price of going to physics rather than to a brochure.

What stacking does not do is sharpen the azimuth pattern of a single antenna. Two Yagis stacked vertically produce the same azimuth pattern as one Yagi, but with the elevation pattern compressed (the elevation 3 dB beamwidth shrinks roughly in proportion to the stacking distance in wavelengths). This is exactly the point on VHF and above where stacking is most popular: tropo-ducted signals come in at 1-3° above the horizon, and a 4-stack with a 5° elevation beamwidth concentrates radiation right where the duct lives, instead of spraying it uselessly up at 15°. On HF, where propagation angles are 5-30° and varying, stacking is less universally productive — a single Yagi at the right height for the favored arrival angle of the moment is often better than a stack stuck with a compressed elevation pattern.

Phased arrays are the alternative geometry: instead of identical antennas in a vertical or horizontal stack, multiple antennas are placed in a 2D layout (corners of a square, points on a circle) and fed with controlled phase shifts to synthesize a directional pattern. The 4-square is the prototypical example — four quarter-wave verticals at the corners of a quarter-wave square, fed in a (0°, -90°, -180°, -90°) phase sequence, produce a cardioid pattern with the maximum in the direction of the lagging vertical. Switch the phasing relays, and the cardioid rotates to face any of the four cardinal directions. The pattern is not as sharp as a multi-element Yagi (3 dB beamwidth is 80-100°, vs. 50-60° for a 4-element 80 m Yagi), but on 160 m and 80 m where building a Yagi is impractical because the boom would be 60+ feet long, a 4-square is the only steerable directional option short of a Beverage farm for receive-only.

The third category, receive arrays, exists because the noise floor on the low bands (160 m, 80 m, 40 m) is dominated by atmospheric and man-made noise pickup, and an antenna with a 25-30 dB front-to-back ratio rejects much of the noise coming from the wrong direction. A 4-square produces 25-30 dB F/B; a properly terminated single beverage gives 15-25 dB F/B over its main lobe; a phased beverage pair pushes F/B to 30+ dB. Crucially, receive arrays are not about gain (most receive arrays have negative gain compared to a dipole; they trade absolute signal level for the noise-rejection that matters). A modern transceiver with a separate receive antenna port (Yaesu FTDX10, FTDX101D, Icom IC-7610, IC-7300 with the optional RX input, Kenwood TS-590SG/TS-890S, Elecraft K3/K4) routes the receive array to the receiver while the main TX antenna stays on the transmitter — the antenna-farm posture is implicit in the rig’s design.

Antenna switching, finally, is the integration layer that ties everything together. A typical contest station has 5-10 antennas at any given moment (a tribander at 70 feet, an 80 m vertical with radials, a 160 m inverted-L, a 4-square for 80 m, a beverage farm for receive, a 144 MHz Yagi for IARU, a 432 MHz Yagi for backup), and the operator has to switch among them in 5-10 seconds while tracking the contest scoring. A relay matrix driven by the rig’s BCD band lines, or by a microHAM Stationmaster + logging-software integration, takes the human factor out of “which antenna am I on now?” and prevents the contest-classic error of calling CQ on the 160 m vertical while the rig is set to 14.230 MHz.

3. Vertical stacking — 2-bay, 3-bay, 4-bay Yagis

Vertical stacking — multiple horizontally-polarized Yagis at different heights on a single tower — is the canonical VHF/UHF gain-recipe. On 6 m, 2 m, 1.25 m, 70 cm, and 23 cm, where boom lengths for high-gain Yagis are manageable (a 9-element 144 MHz Yagi has a 24-foot boom; an 18-element 432 MHz Yagi has a 16-foot boom), stacking 2 or 4 of them produces gain that no single Yagi of any reasonable boom length can match. The math: a 9-element K1FO 2 m Yagi delivers 13.5 dBi in free space; a 4-stack of the same Yagi delivers 18.5 dBi in free space. To match 18.5 dBi with a single Yagi, the boom would need to roughly quadruple to 100+ feet, which is structurally absurd at 2 m wavelengths.

The mounting hardware is the H-frame (for 2-stacks) or K-frame (for 4-stacks). An H-frame is two cross-boom horizontal members on the tower, each carrying one Yagi at the same azimuth — the H comes from the side-view shape of the tower-plus-two-crossbooms. A K-frame is the 4-stack equivalent: four horizontal cross-booms on a single vertical mast, each carrying one Yagi, all at the same azimuth. Commercial H-frame and K-frame hardware comes from Force 12 (now Array Solutions, since the 2019 acquisition), M2 Antenna Systems, and DXEngineering — expect $400-1000 for an H-frame, $1500-3500 for a K-frame, in mid-2026 USD. DIY is feasible with 2-inch Schedule-40 aluminum tube, U-bolts, and saddle clamps; budget $200-400 in materials for either configuration.

Spacing between bays is the engineering crux. The optimum spacing maximizes forward gain by reinforcing the in-phase radiation in the broadside direction while minimizing the secondary lobes that steal energy. For two identical Yagis, the optimum vertical spacing is approximately equal to the half-power beamwidth of the single antenna expressed in wavelengths. For a 5-element 144 MHz Yagi with a 50° elevation beamwidth (single antenna), the optimum 2-stack spacing is approximately 1.0 λ (~7 feet at 144 MHz). For a longer-boom 12-element Yagi with a 28° beamwidth, the optimum spacing is 0.6 λ. Going closer than optimum spacing reduces gain (the antennas couple too tightly); going wider increases secondary lobes (the elevation pattern develops nulls and sidelobes that steal energy from the main lobe). The rules of thumb: 5/8 λ to 1 λ between bays for VHF/UHF Yagis is the sweet spot. Below 0.5 λ the mutual coupling kills the stacking gain; above 1.5 λ the sidelobes eat it.

The phasing harness — the cable network that takes the signal from the radio and divides it equally and in-phase to each Yagi — is where amateur installations most often fail. The most common topology is the electrical-half-wave harness: an N-input, N-output Wilkinson splitter feeds N feedlines of identical electrical length, each terminating at one Yagi’s feedpoint. The electrical-length identity is what guarantees in-phase combination at the splitter; even a 1-foot mismatch between two harness arms at 144 MHz introduces about 15° of phase error, which is enough to lose 0.3-0.5 dB of stacking gain. The trick is electrical length, not physical length: coax has a velocity factor (LMR-400 is 0.85, RG-58 is 0.66, Heliax LDF4-50A is 0.88), and a half-wavelength of coax at 144 MHz is 88 cm of LMR-400 or 68 cm of RG-58. The harness arms must be cut to identical electrical length, which means identical physical length only if they are the same coax type (mixing LMR-400 and RG-58 in the same harness without VF-correcting the lengths is a classic and expensive error). A NanoVNA in time-domain reflectometry mode (Vol 24) lets you measure each harness arm’s electrical length to ±1 cm, which is the precision you want.

The Wilkinson splitter at the splitter end is covered in Vol 18 (Passive splitters). For a stacking harness, the splitter must be matched to the impedance of the parallel-combined Yagi feedlines (two 50 Ω feedlines in parallel present 25 Ω; the splitter is a 2-way 50 Ω-in, 25 Ω-out divider, which is unusual). The more common simplification uses a quarter-wave matching transformer between the splitter output and the parallel-coax bundle, transforming the 50 Ω splitter port to 25 Ω at the antenna side. DXEngineering and Array Solutions sell pre-built stacking harnesses with matched feedline arms and integrated splitters — expect $200-600 for a 2-stack VHF harness, $500-1200 for a 4-stack. DIY with cut-to-length LMR-400 and a Mini-Circuits ZN2PD-04003-S+ or equivalent Wilkinson is $100-200 in parts.

Phase-matching can also be done with phasing-stub harnesses — segments of coax of specific electrical lengths (half-wavelength or one-wavelength) that are added in series with each harness arm to bring all arms to the same total electrical length. This is the W2DU technique and is common on contest-grade installations where the antennas are not equidistant from the splitter. The price of the technique is an extra 0.1-0.3 dB of harness loss per phasing stub; the benefit is the freedom to mount Yagis at different physical distances from the splitter location.

For a 4-stack VHF or UHF installation, the K-frame plus harness plus four Yagis plus a Hy-Gain Ham IV or Yaesu G-1000DXA rotator typically runs $4,000-8,000 in mid-2026 USD for premium-tier (M2 or Force 12) Yagis, or $2,500-5,000 for the budget tier (Cushcraft, Diamond, or Innov). The DIY-everything route with surplus aluminum and Mini-Circuits parts can land at $1,500-2,500.

4. Horizontal stacking — phasing for fixed-direction gain

Horizontal stacking is the lower-frequency cousin of vertical stacking: two or four Yagis side-by-side on a single horizontal cross-boom, all at the same height, fed in phase to produce additional broadside gain. The geometry is most commonly seen on the long-boom HF Yagis — 5- to 7-element 20 m Yagis (50-60 foot booms), or 3-element 40 m Yagis (40-50 foot booms) where adding height-stacking is mechanically impractical because the antennas are simply too big and heavy to support multiple bays vertically.

The K1FO 2-stack 6 m configuration is the canonical example. Two 9-element K1FO Yagis on a 12-foot horizontal cross-boom at 70 feet on the tower, fed in phase by a half-wave-electrical harness, produce 19 dBi in free space (vs. 16 dBi for the single Yagi), with a 25° azimuth beamwidth (vs. 40° for the single). The horizontal spacing for optimum gain is 1.0 λ at 50 MHz — 6 m — so the cross-boom is 12 feet long. The pattern is fixed in azimuth (the stack only points in one direction unless the entire cross-boom rotates), but for fixed-direction work (a beacon, a digital-mode propagation monitor, or a contest single-band entry where you only point at one region) this is acceptable.

The 4-stack 144 MHz EME-class horizontal stack is the next step up. Four 17-element K1FO 2 m Yagis on a 2x2 cross-boom array at 60 feet on the tower deliver 22 dBi in free space — roughly 6 dB over the single 17-element Yagi. The cross-boom geometry is two horizontal cross-members spaced 1 λ vertically (2 m), with two Yagis on each cross-member at 1 λ horizontal spacing. The phasing harness is a 4-way Wilkinson splitter with four matched-length feedline arms. This configuration is at the high end of what amateur 2 m EME operators build, and it delivers reliable signal copy on JT4F and Q65 modes with 1.5 kW PEP into the array; the dish-equivalent equivalent isotropic radiated power is around +85 dBm.

The commercial market for stacked Yagis is dominated by M2 Antenna Systems (the K1FO 6 m, 2 m, and 70 cm reference designs are M2 products), OptiBeam (German manufacturer, premium-tier HF Yagis with optimized stacking ratios), Force 12 / Array Solutions (acquired 2019; the C-3 / C-4 / C-5 series 3-element through 5-element HF Yagis are stacking-friendly), and Cushcraft (mid-tier HF and VHF, lower price point). Expect $5,000-15,000 for the antennas alone for a stacked HF installation, $3,000-8,000 for a stacked VHF EME-class installation, mid-2026 USD.

DIY is more accessible than the commercial price suggests, because the antenna part of a stack is just two of the same antenna — if you’ve already built one 5-element 20 m Yagi from aluminum tubing per Vol 11, building a second is the same exercise. The harness and the rotation/mounting hardware is what eats the budget — a rotator capable of handling 50+ square feet of wind load is $1,200-2,500 (Yaesu G-2800SDX, Hy-Gain T2X), and the K-frame or cross-boom is $400-800 in 2-inch aluminum tubing and U-bolts.

5. Phased verticals — broadside, end-fire, cardioid

Phased verticals — multiple quarter-wave verticals fed with deliberate phase relationships — are the low-band (160 m, 80 m, 40 m) answer to “how do I get directional gain without building a 60-foot-boom Yagi?” The canonical configuration is the 4-square: four λ/4 vertical monopoles (Vol 8) at the corners of a square with sides of λ/4, fed by a four-port phasing controller that imposes specific phase shifts on each vertical. The result is a cardioid radiation pattern with the maximum in one of the four cardinal directions; switching the relay configuration in the controller rotates the cardioid 90°, 180°, or 270° instantly.

The phase relationships in a 4-square are non-obvious. The Christman feed (the standard amateur 4-square topology, originated by Al Christman K3LC in 1990) uses (0°, -90°, -180°, -90°) — front vertical at 0°, two side verticals at -90°, back vertical at -180°. This produces a cardioid with a 25-30 dB front-to-back ratio, 80-90° azimuth beamwidth, and roughly 3 dB gain over a single vertical in the favored direction. The feed currents are not equal in magnitude — the front and back verticals carry roughly 1.0 unit of current each; the two side verticals carry 1.4 units each (the difference is what tilts the pattern from omnidirectional toward the cardioid). The phasing-and-amplitude relationships are imposed by the Christman network — a combination of phasing coax, hybrid transformers, and a dummy-load termination that absorbs the unwanted-phase components.

The Lewallen feed (W7EL, Roy Lewallen) is an alternative 4-square topology that uses (0°, -111°, -180°, -111°) and produces a slightly different pattern with the same general performance. The Lewallen network is simpler than the Christman in some implementations (it uses fewer phasing components) but has higher feedline-length sensitivity. Most commercial 4-square controllers in mid-2026 implement the Christman feed.

The commercial 4-square controller market has three serious players. Comtek 4-Square Systems (Hi-Z Antennas) sells the ACB-4 controller for $800-1,200 with the matching feed network — a quality entry point for 80 m or 40 m use. Array Solutions sells the StackMatch IV and similar phasing controllers in the $1,200-2,000 range — heavier-duty relays, more weather-protection, computer-control interface for contest stations. DXEngineering sells the 4-square phasing controllers with full kit integration including the four verticals, radial system parts, and Christman feed network — $2,000-4,000 for a complete kit, mid-2026 USD. The DXEngineering DXE-4S-1 controller alone is approximately $1,500.

DIY 4-square controllers are a popular winter project. The phasing network requires four λ/4-electrical coax stubs (one per vertical), a 75 Ω quarter-wave matching section to transform the 25 Ω parallel-combined feed-impedance to 50 Ω, a relay-switched feed network to flip the (0°, -90°, -180°, -90°) sequence around to any of the four cardinal directions, and a dummy-load termination to absorb the unwanted-phase port output. A PIC microcontroller (PIC16F84 in the 2002 W2DU design, or a modern ATmega328P) drives four high-current relays (Tyco T9AS5D12-12 or equivalent, 12 VDC coil, 30 A contacts) that select which vertical is which in the rotation. BOM is $300-500 in mid-2026 parts, plus the four verticals themselves (commercial Hustler 6-BTV or DXE-160VA-1 at $300-500 each, or homemade with aluminum tubing and a radial field per Vol 8).

The radial system for a 4-square is the unglamorous heavy work. Each vertical needs its own radial field — for 80 m, the standard is 60-120 radials of λ/4 length each (~65 feet), laid on the ground and connected at the vertical’s base. The four radial fields can share radials at the boundaries between verticals, but in practice each is built as a standalone radial field with the boundary-region radials terminated independently. Total wire requirement for a 4-square 80 m installation is approximately 240 radials × 65 feet = 15,600 feet of #14 AWG bare copper wire — that’s roughly 50 lbs of wire and $600-900 in mid-2026 USD. The radial field installation work (digging shallow trenches, laying wire, ground-stake connections) takes 40-80 hours of labor depending on yard conditions.

Where the 4-square wins is exactly where a single-Yagi-on-a-tower loses: 80 m and 160 m DX work, where building a 3-element 80 m Yagi would require a 50-foot boom (~$8,000 in commercial hardware, plus a $2,500 heavy rotator and a $4,000 tower upgrade), and where the propagation angles (5-15° elevation) and noise environment make a steerable cardioid worth more than 3 dB of additional gain. For a contest operator on 40 m where the band noise is high and the closely-packed CW contest signal density is dense, a 4-square that nulls the European interference toward the Caribbean is worth multiple QSO multipliers per contest.

6. Beverage arrays for low-band receive

A single beverage antenna — a horizontal wire of 1-4 wavelengths terminated in 470 Ω at the far end (Vol 15 (Receive-only loops)) — is the canonical low-band receive-only antenna. It has negative gain compared to a dipole (typically -10 to -20 dBi), but its 25-35 dB front-to-back ratio and ~50° azimuth beamwidth make it the right tool for hearing weak signals through the high atmospheric noise floor of 160 m and 80 m. A beverage array — two or more beverages combined with phasing — pushes the performance further: phased pairs produce 30-40 dB F/B and 60° azimuth beamwidth with the same -10 dBi gain, and the array can be steered electronically by changing the phasing in the controller.

The basic phased-beverage-pair topology uses two parallel beverages spaced λ/2 perpendicular to their length, fed by a phasing harness that introduces a controlled phase shift between them. With the two beverages at 0° and -90° phase, the combined pattern is cardioid with the maximum in the forward direction; switching the phasing to (0°, +90°) flips the pattern to point in the opposite direction. The Hi-Z Antennas NCC-2 receive controller ($1,500-3,000 in mid-2026 USD) is the canonical commercial implementation — a two-channel receive amplifier with phase-and-amplitude control over a beverage pair, plus a relay-switched termination network. The DXEngineering NCC-2 / DXE-NCC-2P is the same product under DXE branding. The output of the NCC-2 is a single coax feed that goes to the receiver’s separate-RX-antenna input port.

Multiple beverages aimed at different azimuths, switched by a relay matrix, is the next level. A typical 8-beverage farm has beverages at 0°, 45°, 90°, 135°, 180°, 225°, 270°, 315° azimuths (covering 360° in 45° increments), all terminated in 470 Ω at the far ends, all feeding their respective coax cables back to a shack-mounted 8-way relay-switched selector. The DXEngineering BFS-1 Beverage-Farm Selector is the canonical product at $600-900 in mid-2026 USD; the relay box is a sealed 8-position SPDT (or SP8T) configuration driven by a manual selector knob or a remote-control panel from the operating position. The total beverage-farm installation typically runs $2,000-4,000 in wire, termination resistors, baluns, and the selector hardware — plus the real estate to lay 8 beverages of 600-2,000 feet each, which is the gating constraint.

The receive-port-equipped transceiver is non-negotiable for beverage arrays. The Yaesu FTDX10/FTDX101D, Icom IC-7610/IC-7300 (with the optional ANT input enabled), Kenwood TS-590SG/TS-890S, and Elecraft K3/K4 all have a separate-RX-antenna input that routes only to the receiver (the transmit signal stays on the main TX antenna). Some transceivers route the RX-antenna input through a relay that disconnects on transmit to protect the LNA in the beverage-controller amplifier — others rely on the operator manually selecting between the antennas. Without a separate RX-antenna port, a beverage array is impossible to use practically because the antenna would have to be physically disconnected and reconnected between transmit and receive every QSO.

The beverage-array performance pays off most clearly on 160 m and 80 m where the noise floor in a typical northeast-US summer evening is -85 to -95 dBm in a 500 Hz bandwidth (from a combination of atmospheric noise, distant thunderstorms, and power-line noise). A single beverage drops the noise floor by 15-20 dB in the array’s null direction; a phased beverage pair drops it by 25-30 dB. A typical European DX signal that registers at -110 dBm on a single inverted-L is below the noise floor on that antenna but clearly audible (+10 to +15 dB SNR) on the receive array. This is what makes 160 m DX possible in noisy locations.

7. Switching matrices — Top Ten Devices, Array Solutions

A switching matrix is an N-input, M-output RF relay network that lets the operator (or a computer) route signals between multiple antennas, multiple radios, and multiple operating positions in arbitrary combinations. The simplest is a 2-in 1-out antenna A/B switch ($30-60 manual, $80-150 remote-controlled); the most complex is a fully-meshed N×M crossbar that lets any of N antennas connect to any of M radios while maintaining isolation between simultaneously active paths (Array Solutions Six-Pak — 6 antennas to 2 radios, $1,200-1,800 in mid-2026 USD).

Top Ten Devices is the long-time amateur specialist for high-power switching matrices. The Top Ten A/B (2-position), A/B/C (3-position), A/B/C/D (4-position), and the 6-position antenna switches use vacuum relays (Jennings RA-30 or equivalent, 5 kV / 30 A contacts) and are rated for 1.5 kW PEP continuous, 3 kW PEP intermittent. Insertion loss is 0.05-0.10 dB per switch position, isolation is 60+ dB between unselected ports, SWR through any selected path is <1.2:1 from DC to 30 MHz. Pricing in mid-2026 USD: $150-200 for the 2-position, $200-280 for the 4-position, $350-450 for the 6-position. Remote-control accessories (RS-232 or USB serial control) are $80-150 additional.

Array Solutions sells the Six-Pak — a 6×2 matrix that lets 6 antennas connect to either of 2 radios in any combination, with crossover protection that prevents the same antenna from being connected to both radios simultaneously. Price is $1,200-1,800 in mid-2026 USD. The Six-Pak is the contest-station workhorse for SO2R (Single-Operator, Two-Radios) where the operator runs CQ on radio 1 with one antenna while tuning the second VFO on radio 2 with a different antenna in parallel.

DXEngineering sells the DXE-RBS-1B (Remote Antenna Switch) at $400-600, with 6 antenna ports and remote-control inputs that accept BCD band-data from the rig’s accessory port. Combined with the DXE-AS-1 antenna selector, this gives full automatic band-switching (rig sends band code; matrix selects the antenna pre-configured for that band).

MFJ Enterprises sells the MFJ-4716 6-position antenna switch at $150-200 (manual rotary), and the MFJ-4724 remote-controlled 4-position switch at $200-300. MFJ’s reputation for build quality has been uneven (the company shut down in 2023, but inventory still moves through dealers in mid-2026); the switches work, but expect to inspect and service contacts annually if used heavily.

Ameritron sells the RCS-12 — a single-coax remote-controlled antenna switch (one coax cable from the shack to the switch box carries both the RF signal and the DC control voltage for the relays). Price is $700-900 in mid-2026 USD. The single-coax topology is ideal for tower-mounted switches where running a separate control cable up the tower is impractical.

The relay technology behind the switches matters for high-power use. Vacuum relays (Jennings RA-30, Gigavac G81, Kilovac K81) have very low contact resistance (<5 mΩ), very low insertion loss (<0.05 dB DC-30 MHz), and very high isolation (70+ dB) — but they are rated for ~1 million switching cycles at full RF power, after which the contacts pit and the loss increases. A serious contest station may exceed 1 million switches in a single year (each QSO involves a band-change-and-switch operation; 1,000-1,500 QSOs per contest weekend × 50 contests per year × multiple switches per QSO). DC sealed-coil relays (Tyco T9AS, Omron G2RL) are cheap ($5-15) and infinite-cycle-rated, but only handle 100-500 W PEP without contact erosion. PIN-diode switches (covered in Vol 30 for multi-radio sharing) are silicon-based, infinite-cycle, no moving parts — but limited to ~100 W PEP and have higher insertion loss (0.3-0.8 dB) at HF frequencies.

For a typical Amateur Extra-class contest station running 1.5 kW PEP on HF, the vacuum-relay path is the right choice. Plan to replace contacts every 3-5 years of heavy use, or have a spare relay set on hand. For a 100 W casual station, DC sealed-coil relays are fine and far cheaper.

8. Automatic antenna selectors per band

The natural extension of the manual switching matrix is the automatic band-selector: a controller that reads the rig’s band-data output (a BCD-encoded band number on the rig’s accessory port) and automatically selects the antenna pre-configured for that band. The operator changes bands on the rig; the antenna switches automatically without any human intervention.

The band-data interface varies by manufacturer. Yaesu rigs (FTDX10, FTDX101D, FT-991A) output a 4-bit BCD band code on the rear-panel accessory connector — band 1 = 160 m, band 2 = 80 m, band 3 = 40 m, etc. Icom rigs (IC-7300, IC-7610, IC-7851) output a similar 4-bit BCD plus a “transmit inhibit” line. Kenwood rigs (TS-590SG, TS-890S) output BCD band data on the ACC2 connector. Elecraft (K3, K4) outputs band data on the AUX port plus a more sophisticated CI-V-style serial interface.

The band-decoder controller is the device that reads the BCD band data and translates it to a relay-control signal for the antenna switch. Commercial options:

DX Engineering DXE-RBS-1B-CTL ($300-400) — a band-decoder that interfaces directly with the DXE-RBS-1B remote antenna switch. Reads 4-bit BCD from any Yaesu, Icom, or Kenwood rig; switches the appropriate antenna automatically.
Top Ten Devices Band Decoder ($250-350) — interfaces with Top Ten antenna switches and provides BCD-to-relay translation plus an over-power lockout.
microHAM Stationmaster ($1,500-2,500) — full station-automation system: reads band data, controls antennas, controls amplifier inputs/outputs, integrates with logging software (N1MM Logger, Win-Test, DXLab), and supports up to 8 antenna ports and 4 radio ports in arbitrary combinations.
Logikey K-3 ($1,200-2,000) — competing station-automation system with similar feature set to Stationmaster; popular in contest circles.

The microHAM Stationmaster is the gold standard for serious contest stations. Plug it into the station LAN, run the configuration software, and tell it which antenna serves each band on each radio. Plug the BCD band-data outputs from each radio into the Stationmaster. Plug the antenna switch and amplifier into the Stationmaster’s relay outputs. Configure the logging software (N1MM Logger has full Stationmaster integration since version 1.0.7000 in 2017) to tell the Stationmaster what to do at each band change. The result: operator changes bands in N1MM, Stationmaster automatically switches the antenna, the amplifier input matching, and the amplifier output network all to the correct configuration in less than 200 ms. The error rate (wrong antenna for the wrong band) drops to near zero, and the contest QSO rate goes up because the operator never has to think about the antenna farm.

Integration with logging software is where this gets really powerful. N1MM Logger has the concept of “antenna definitions” — a database of which antenna serves which band, which beam direction the antenna is pointed (for rotatable antennas), and which receive antenna pairs with which transmit antenna. The Stationmaster reads this from N1MM in real-time and executes the antenna selection automatically. For an operator running a multi-band contest entry (CQWW, ARRL DX, IARU), this transforms the experience from “manually tracking 8 antennas across 6 bands” to “just operating the radio.”

9. Coupling networks — Christman, Lewallen, hybrid

The Christman 4-square feed network, introduced earlier in §5, is the workhorse phasing topology for amateur 4-square installations on 160 m, 80 m, and 40 m. The Christman feed uses a combination of phasing coax (a half-wavelength of coax provides a 180° phase shift; a quarter-wavelength provides a 90° shift; a three-eighths-wavelength provides a 135° shift) plus a 75 Ω quarter-wave matching transformer and a dummy-load termination that absorbs the unwanted-phase port output.

The phasing components in a Christman 4-square (using Belden 9913 coax, VF 0.84, for 80 m operation at 3.6 MHz):

Component	Physical length	Electrical length	Purpose
Phasing line A (front vertical to feed point)	1.0 λ × 0.84 = 70 ft	1.0 λ	Reference 0° phase
Phasing line B (left side vertical to feed point)	0.875 λ × 0.84 = 61 ft	0.875 λ	-45° phase shift from reference
Phasing line C (back vertical to feed point)	0.5 λ × 0.84 = 35 ft	0.5 λ	-180° phase shift from reference
Phasing line D (right side vertical to feed point)	0.875 λ × 0.84 = 61 ft	0.875 λ	-45° phase shift from reference
75 Ω matching transformer (quarter-wave)	0.25 λ × 0.66 = 14 ft RG-11	0.25 λ	Matches 25 Ω parallel feed to 50 Ω
Dummy load termination	50 Ω, 100 W	—	Absorbs unwanted-phase port output

Phase shifts in the Christman network are not arbitrary — they are set by the relationship between phasing-line length, antenna driving impedance, and the relay-network configuration. The 4-square produces a cardioid pattern when the (0°, -90°, -180°, -90°) phase sequence is imposed; that sequence requires the asymmetric (0°, -45°, -180°, -45°) phasing lines plus the Christman network’s internal phase shifts. The math is detailed in Al Christman’s QST article (March 1990 issue) and in the ARRL Antenna Book Chapter 11.

The Lewallen feed (W7EL, Roy Lewallen, 1995) is an alternative 4-square topology that uses (0°, -111°, -180°, -111°) phasing and a hybrid-coupler-based feed network. The Lewallen network is mechanically simpler (fewer phasing-coax stubs) but more sensitive to feedline-length errors than the Christman; it is less common in commercial 4-square controllers because the Christman has slightly better F/B and slightly easier tuning.

Hybrid coupler-based phasing is the third major topology, used primarily in commercial broadcast and military 4-square installations. A hybrid coupler (a 90° hybrid is a 4-port device that splits input power 50/50 between two outputs with a 90° phase difference; a 180° hybrid splits 50/50 with 180° phase difference) lets the designer impose precise, frequency-independent phase relationships between vertical pairs. The Wilkinson splitter (90° hybrid topology, see Vol 18) is the canonical hybrid coupler. Hybrid-coupler-based 4-square networks have wider bandwidth (the phasing stays correct across the full amateur band) but cost more in parts and require more sophisticated test equipment to verify.

For amateur 4-square installations on a single band, the Christman feed is the standard choice. For 80 m + 40 m dual-band 4-square (where the network must work on both bands), the hybrid-coupler approach is preferred because it has wider bandwidth and lower frequency-sensitivity.

10. SO2R and multi-Op station design

SO2R (Single-Operator, Two-Radios) is a contest-operating mode where one operator runs two radios in parallel: typically one radio calls CQ on a “run” frequency while the other radio tunes and answers other stations’ CQs on a “search and pounce” frequency. The two radios connect to two different antennas (or two ports of a multi-port matrix). The operator wears a headset that switches audio between the two radios — typically left ear is radio 1, right ear is radio 2, with footswitches or keyboard shortcuts to focus attention.

For SO2R, the antenna posture is critical. The two radios must use antennas with sufficient isolation that radio 1’s transmit signal doesn’t desensitize radio 2’s receiver (typically 50-70 dB of inter-antenna isolation is required, which means the antennas must be at least 100 feet apart and pointed in different directions, or there must be a band-pass filter on each receiver to reject the other radio’s transmit frequency). The Array Solutions Six-Pak is the canonical SO2R switching matrix — it lets the operator assign any of 6 antennas to either of 2 radios, with crossover protection that prevents both radios from connecting to the same antenna.

Multi-Op (multiple operators, multiple radios, multiple antennas) is the contest mode of the major contest stations (CQWW M/M, ARRL DX M/M, IARU M/M categories). A multi-multi station has 4-8 operators each running their own radio with their own antenna, all hosted in a single building with shared infrastructure (power, cooling, networking, logging software). The antenna farm has 10-30 antennas across 8-12 bands; the switching matrix is N×M crossbar (e.g., 16 antennas × 4 radios = 64 possible combinations) with logging-software integration for automatic configuration.

Antenna-switching pulse rates during contests are 5-10 second band-changes — the operator changes bands, the antenna matrix switches, the amplifier retunes, and the operator continues operating. The matrix relays (and the amplifier band-switch contacts) take the brunt of this — a competitive contest weekend may exercise the antenna matrix 5,000-10,000 times, which is why vacuum relays are mandatory at full power and why mechanical maintenance is built into the contest-station calendar.

The receive-vs-transmit antenna split is the key tactical decision on low-band contest entries. On 160 m and 80 m, the optimal configuration is a transmit antenna (a vertical with a serious radial field, or an inverted-L with a top-loading capacitive hat) for the high-power transmit signal, and a separate receive antenna (a 4-square, beverage array, K9AY, or Hi-Z circle) for the receive signal. The TX antenna optimizes for radiation efficiency; the RX antenna optimizes for noise rejection. The two are routed independently — TX antenna goes to the transmitter port, RX antenna goes to the separate-RX-antenna input on the receiver. This is the configuration that turns 160 m DX from “hopeless on a quiet evening” into “regular contacts to Europe and the Pacific.”

11. DIY build — a 2-element broadside-phased vertical array

The 2-element broadside-phased vertical array is the entry-level phased-vertical project — two quarter-wave verticals spaced 1/2 wavelength apart, fed in phase, with a fixed broadside-pattern. The pattern is bidirectional (forward and backward gain are equal, perpendicular to the line connecting the two verticals), with about 3 dB gain over a single vertical and 50-60° azimuth beamwidth in each lobe.

BOM (mid-2026 USD):

Item	Source	Cost	Notes
2× DX Engineering DXE-160VA-1 vertical, 80 m	DXEngineering	$620	Or DIY aluminum verticals per Vol 8 — ~$200 in parts each
60× radial, λ/4 80 m (~65 ft), #14 AWG bare copper	Wire House Direct	$180	240 radials total if you go to 60-radial fields per vertical
2× Comtek phasing-coax stub, λ/2 80 m	DXEngineering or DIY	$80	Or cut 0.5 λ × 0.84 = 35 ft of LMR-400
50 Ω 100 W dummy-load termination (only needed for switching, not for fixed-broadside)	Comtek or DIY	—	Skip for fixed-direction broadside
2-position remote-switching relay box (for direction reversal)	DXE-RBS-1B or DIY	$150-400	Skip for fixed-broadside; needed for direction-flip
RG-213 feedline from antenna pair to shack (~100 ft)	DXEngineering or Bury-Flex	$90	Or LMR-400 if you want lower loss
Misc: coax connectors, weatherproofing, hardware	—	$80
Total		$1,200-1,400

Setup procedure:

Site selection. Two verticals at 1/2 wavelength spacing (132 feet apart for 3.6 MHz) require a clear east-west line (for a broadside pattern aimed north-south) or north-south line (for an east-west broadside pattern). The verticals need to be in the clear of large metal objects (other antennas, gutters, fences) for at least 20 feet in all directions.
Install the radial fields. Each vertical gets 60 radials of 65 feet each. Lay the radials radially from the vertical’s base, on the ground surface or buried just below grade. Bond all radials at the vertical’s base with a #6 AWG copper bus, then bond the bus to the vertical’s grounding lug.
Install the verticals. Mount each vertical on a 4-foot insulated mast (PVC or fiberglass) above the radial field. Attach the feedline at the vertical’s base.
Cut the phasing coax stubs. Each stub is 1/2 wavelength of LMR-400 at 3.6 MHz: 1/2 × 80 m × 0.84 (VF of LMR-400) = 35 feet. Cut precisely — use a NanoVNA in TDR mode (Vol 24) to verify electrical length is within ±1% of 35 feet.
Run the feedlines. From each vertical’s feedpoint, run 35 feet of LMR-400 phasing stub to a central T-junction. From the T-junction, run RG-213 or LMR-400 back to the shack (~100 feet typical).
Tune the system. Connect the shack-end of the feedline to an antenna analyzer or NanoVNA. Sweep the band and verify the impedance at the feed point is approximately 25-30 Ω (resistive, with no reactance) at the operating frequency. Adjust the vertical lengths (trim from the top) to bring the impedance to the target.
Test the pattern. Use a portable signal source at 1-2 km away in the broadside direction; measure received signal level. Move the signal source to a position 90° off-axis from the broadside direction (in the null direction); verify the received signal drops by 15-25 dB. This confirms the array pattern.

The result is a fixed bidirectional broadside pattern with about 3 dB gain over a single vertical, deployed for a one-band installation. For direction control (broadside, end-fire-forward, end-fire-back), add the relay-switching network to flip the phasing between the two verticals from 0° (broadside) to ±90° (end-fire). Total cost increases by $200-400 for the relay box and the additional phasing stubs.

12. Commercial buys

Sorted by price tier (USD, mid-2026):

Budget tier (~$150-500):

Top Ten Devices A/B 2-position antenna switch — $150-200
Top Ten Devices A/B/C/D 4-position antenna switch — $200-280
MFJ-4716 6-position antenna switch (manual rotary) — $150-200
MFJ-4724 remote-controlled 4-position switch — $200-300
Top Ten Devices 6-position vacuum-relay switch — $350-450
DXEngineering DXE-RBS-1B 6-position remote switch — $400-500

Mid tier (~$600-1,800):

DXEngineering DXE-RBS-1B-CTL band-decoder controller — $300-400 (paired with DXE-RBS-1B switch)
Top Ten Devices Band Decoder — $250-350
Ameritron RCS-12 single-coax remote antenna switch — $700-900
Comtek 4-Square Controller (Hi-Z Antennas ACB-4) — $800-1,200
Array Solutions Six-Pak 6×2 antenna matrix — $1,200-1,800
DXEngineering NCC-2 receive antenna controller — $1,500-2,500
Hi-Z 4 Pro receive 4-square array (complete kit) — $1,200-1,800

Premium tier (~$2,000-15,000):

DXEngineering DXE-4S-1 4-square phasing controller — $1,500-2,500 (controller only); $2,500-4,500 (complete 4-square kit with verticals)
microHAM Stationmaster station-automation system — $1,500-2,500
Logikey K-3 station-automation system — $1,200-2,000
DXEngineering 9-circle complete receive-array kit — $5,000-8,000
Array Solutions custom 8-circle complete kit — $7,000-12,000
K-frame + 4× M2 17-element 144 MHz Yagis complete 4-stack EME installation — $8,000-15,000
Force 12 / Array Solutions 5-element 20 m stacked Yagi system (2-stack with rotator) — $8,000-15,000

Stacking-harness components:

DXEngineering pre-built 2-stack VHF harness — $200-600
DXEngineering pre-built 4-stack VHF harness — $500-1,200
Custom Wilkinson splitter (Mini-Circuits ZN2PD-04003-S+) — $40-80 for the part; $100-200 for a packaged assembly with connectors

Tower-stacking hardware:

Rohn 25G tower section, 10-foot — $250-400
Rohn 45G tower section, 10-foot — $450-700
US Towers MA-40 motorized crank-up — $4,000-6,000 (40 ft)
US Towers MA-770MDP motorized crank-up — $8,000-12,000 (70 ft, 50 sq ft wind load)
Yaesu G-450 light-duty rotator — $300-400 (6 sq ft wind load)
Yaesu G-800 medium-duty rotator — $550-750 (12 sq ft)
Yaesu G-1000DXA heavy-duty rotator — $800-1,200 (22 sq ft)
Yaesu G-2800SDX serious-stack rotator — $1,500-2,500 (50 sq ft, 35 sq ft pointing accuracy)
Hy-Gain Ham IV — $400-550 (22 sq ft, the contest-station classic)
Hy-Gain T2X — $700-900 (35 sq ft)
Pro.Sis.Tel PST-2051D / PST-3300 — $1,500-3,000 (heavy stack and large arrays)
Yaesu G-5500 azimuth-elevation rotator (for EME) — $1,000-1,500
AzEl controller for G-5500 — $200-400

13. Common gotchas and myths

“More antennas = more signal.” False. Antennas combined out of phase produce nulls in the favored direction, not gain. A 2-stack with a 180° phasing error in the harness has less gain than a single antenna, and the pattern has a deep null where the gain should be maximum. The 2.5-3 dB stacking gain depends on the antennas being fed in phase at the correct spacing. Phasing errors of >15° produce measurable losses; phasing errors of >45° produce gain inversions.

“Stacking 2 Yagis doubles the gain.” Wrong. The doubled quantity is aperture, which converts to +3 dB of power gain, not a doubling of dB. A single 13.5 dBi Yagi stacked into a 2-stack delivers 16.5 dBi (free space), not 27 dBi. The “doubles the gain” interpretation confuses dB with raw power and is the #1 misunderstanding in amateur stacking discussions.

“Beverage arrays are easy DIY.” Partially true. A single beverage is straightforward — wire, two PVC supports, a 470 Ω termination resistor, a 9:1 unun feeding the receiver. A phased beverage pair with a controller is much harder — the controller requires precise phasing-coax cutting, an active receive amplifier, a phasing network, and a relay-switched termination. The DXE NCC-2 controller exists at $1,500-3,000 precisely because the engineering isn’t trivial. DIY a single beverage; buy the controller for a beverage array.

“Vacuum relays last forever.” Wrong. Vacuum relays are rated for approximately 1 million switching cycles at full RF power. A serious contest station with 5,000-10,000 switches per contest weekend over 50 contests per year may exceed this in a single season. Plan to replace contacts every 3-5 years of heavy use. The failure mode is contact pitting that increases insertion loss; the symptom is a 1-2 dB hot-spot at the switch position when the contacts have degraded.

“Tower stacks are stable in wind because they have multiple anchors.” Actually false. Stacked Yagis amplify wind loads through resonance — the 2-stack and 4-stack configurations have specific wind-loading harmonics that can shake the tower at frequencies well below the static wind load. Tower analysis per TIA-222 (the successor to EIA-222) is non-optional for 4+ Yagi installations. Skipping the analysis is how tower collapses happen.

“EME is amateur-bands only.” True for amateur Moon-bounce EME (144 MHz, 432 MHz, 1296 MHz, 2.3 GHz, 5.7 GHz, 10 GHz), but commercial and military operations have done Moon-bounce on K-band (10-40 GHz) since the 1950s — and modern academic radar-astronomy uses Moon-bounce as a calibration reference for planetary-radar systems. The Arecibo radio telescope (collapsed in 2020) did Moon-bounce at S-band for years as part of its solar-system radar program.

“You need a separate beverage for every direction.” Partially true. A traditional beverage farm has 8-12 beverages aimed at different azimuths, switched by a manual selector. But a phased-beverage pair with the DXE NCC-2 controller can be electronically steered through ±60° of azimuth — so 4-6 beverage pairs with controllers can cover 360° with electronic steering, vs. 8-12 dedicated single beverages.

“A 4-square is omnidirectional when all phases are equal.” False. A 4-square in the “omnidirectional” position (all four verticals fed in phase, equal amplitude) is actually a wide-azimuth pattern with a 6 dB gain peak in two opposing directions and partial nulls in the orthogonal directions — not truly omnidirectional. True omnidirectional pattern requires only a single vertical; the 4-square’s value is its directional cardioid pattern, not any omnidirectional mode.

14. Resources

ARRL Antenna Book, Chapter 11 (Multi-Element Arrays) — the canonical amateur reference; covers stacking, phased verticals, the Christman 4-square, beverage arrays, switching matrices.
ARRL Antenna Book, Chapter 22 (Towers and Structures) — mechanical engineering of tower-stacking, EIA-222 / TIA-222 load analysis, guy-wire geometry.
VHF/UHF/Microwave Experimenter’s Handbook (ARRL) — the canonical reference for VHF/UHF stacking, EME, and high-gain-array work.
WSJT-X manual (K1JT, Joe Taylor) — for the EME-mode protocols (JT4F, Q65) that justify the stacking expense.
ON4UN’s Low-Band DXing (4th edition or later) — the bible for 160 m / 80 m antenna farms, phased verticals, beverage arrays, receive arrays. Indispensable for any serious low-band operation.
Al Christman K3LC “The Christman 4-Square” (QST, March 1990) — the original paper that defined the standard amateur 4-square topology.
Roy Lewallen W7EL “EZNEC Pro+ Documentation” and 4-square modeling files — the alternative Lewallen feed topology and modeling references.
DXEngineering product catalog — the most comprehensive single source for switching matrices, phasing controllers, 4-square kits, beverage selectors, stacking harnesses.
Array Solutions product catalog — competing product line for switching matrices, hybrid couplers, Six-Pak SO2R matrices.
Force 12 / Array Solutions product line — premium stacked Yagi systems for HF and VHF.
M2 Antenna Systems product line — premium VHF/UHF Yagis for stacking and EME.
KrakenSDR documentation (KrakenRF.com) — the multi-channel SDR direction-finding platform mentioned in §5; 5× RTL-SDR with shared clock, MUSIC algorithm for direction-finding, $250-300 in mid-2026 USD.
N1MM Logger documentation (N1MM+, n1mmwp.hamdocs.com) — the contest-logging software with full microHAM Stationmaster integration; the antenna-automation reference.
microHAM Stationmaster documentation — the station-automation gold standard for contest and serious-operation antenna farms.
TIA-222 (Telecommunications Industry Association) tower-structural standard — the successor to EIA-222; mandatory reading for any tower hosting 4+ Yagis or a 4-square installation.
OET-65 calculator (FCC) — required for RF-safety analysis on stacking installations; the calculation is straightforward but the documentation is critical for FCC compliance (Vol 31).

For grounding the antenna farm, see Vol 20 (Grounding). For mounting and tower hardware, see Vol 21 (Mounting, masts). For modeling stack patterns before construction, see Vol 28 (Antenna modeling). For sharing antennas across multiple radios (the inverse problem from this volume), see Vol 30 (Multi-radio shared). For the regulatory envelope (power limits, tower lighting, FAA notification), see Vol 31 (Regulatory). For the BALUN choices in the phasing harness, see Vol 16 (BALUNs). For the Wilkinson splitter at the stacking harness’s input, see Vol 18 (Passive splitters). For active distribution amplifiers in receive-array implementations, see Vol 19 (Active splitters). For full-wave-loop arrays as an alternative geometry to dipole stacking, see Vol 14 (Transmitting loops).

Contents