Chapter 2, Section 4

Inside the Tuning House: Base Insulators, ATUs & the Austin Transformer

Part of AM Radio Facts and Trivia: The AM Broadcast Band in the USA. Updated November 15, 2026.

Every directional array on this site is really a row of little buildings, one crouched at the foot of each tower. Broadcasters call them tuning houses, ATU huts, or, universally, "doghouses." They are where the elegant theory of a phased array meets a few hundred amps of RF current and a tower sitting at thousands of volts above ground.

The base insulator: why the tower floats. A series-fed AM tower doesn't touch the earth electrically. It stands on a single heavy porcelain or steatite base insulator, often no bigger than a stack of dinner plates, that carries the entire dead weight of the structure while keeping it isolated from ground. The whole tower is the "hot" conductor of the antenna, driven against the buried ground radial system. On a tall radiator the RF voltage at the base can run into the thousands of volts, which is why the base insulator sits inside a fenced, interlocked enclosure and why the guy wires are chopped into harmless lengths by strain insulators (the egg-shaped porcelain links old-timers call "Johnny balls") so the guys can't re-radiate and wreck the pattern.

The ATU: matching and phasing. Inside the doghouse is the Antenna Tuning Unit (sometimes Antenna Coupling Unit). At minimum it transforms the tower's base impedance, which is rarely a tidy 50 ohms, to match the transmission line. In a directional array it does more: working with the phasor back in the transmitter building, it sets the phase and amplitude of the current in that particular tower relative to the others. The hardware is gloriously physical: tapped roller inductors the size of paint cans, vacuum or mica transmitting capacitors, copper strap instead of wire, and an RF ammeter on the wall reading the base current. Adjusting one is a slow conversation between the tuning house and the antenna monitor (see The Antenna Monitor).

The Austin transformer: lighting a tower that's electrically hot. Here's a problem most listeners never think about: the FAA wants red obstruction lights burning at the top of that tower, but the tower is floating at high RF potential. Run an ordinary mains wire up it and you'd short the antenna to ground through the power company. The classic solution is the Austin ring transformer: an air-gapped toroidal transformer whose primary winding sits on the grounded side and whose secondary rides on the tower, the two halves separated by a deliberate physical gap. Power crosses the gap by magnetic coupling; RF cannot. For decades that gapped ring at the base was how nearly every insulated tower in America kept its lights lit. Newer sites increasingly skip it in favor of fiber-fed or photovoltaic lighting controllers that need no copper path across the insulator at all, but walk an older site and you'll find the Austin ring still doing its quiet trick.

Sources: Kintronic Labs (ATU / phasing literature) · Austin Insulator Co. · NAB Engineering Handbook · ARRL Antenna Book (MF verticals).

THE BASE INSULATOR: WHERE THE HOT TOWER MEETS THE EARTH concrete pier ATU (tuning house) ⚡ ground radials (buried) Everything above the insulator is HOT: the whole tower is the antenna, carrying full RF voltage Porcelain compression insulator carries the tower's full weight in compression while blocking RF to ground; ribs lengthen the leakage path in rain and grime Ball gap lightning jumps the gap to ground instead of through the insulator Feed strap copper strap from the ATU drives the tower base Ground strap pier hardware bonds to the radials

The whole tower stands on a ceramic foot. In a series-fed AM tower the entire structure is the radiating element, so it can't touch ground: hundreds of tons of steel rest on a porcelain compression insulator the size of a milk can. Porcelain is enormously strong in compression (the one shown handles the tower's full weight with a wide safety margin) while blocking the RF from leaking to the pier. The ribbed sheds stretch the surface leakage path so rain and dirt can't creep a short across it, and the ball gap alongside gives lightning a deliberately easier route to ground than punching through the porcelain. Guy anchors get the same treatment with strain ("Johnny ball") insulators; see the story text.

Drawing 2.4-1

INSIDE THE ATU: A T-NETWORK AT 1000 kHz tuning house ("doghouse") from transmitter 50 Ω coax L1 +j42.4 Ω → 6.75 µH C −j42.4 Ω → 3,750 pF L2 +j21.4 Ω → 3.4 µH static drain choke (high Xₗ) feed strap tower base Zₐₙₜ = 36 + j21 Ω 90° tower (234 ft @ 1000 kHz) THE ARITHMETIC: 1000 kHz, 90° T-NETWORK Match 50 Ω line to 36 Ω tower: every leg gets |X| = √(50 × 36) = 42.4 Ω L1 = 42.4 / (2π × 1 MHz) = 6.75 µH C = 1 / (2π × 1 MHz × 42.4) = 3,750 pF Output leg must total +j42.4 Ω, but the tower already brings +j21 of its own, so L2 only supplies the difference: +j21.4 Ω → 3.4 µH. The tower's reactance is absorbed into the network.

Three parts, one job. The classic ATU is a T-network (coil in, capacitor down, coil out) that transforms whatever impedance the tower presents into the clean 50 Ω the transmission line wants, while shifting phase by a designed amount (here 90°, the standard choice, and in a directional array, that designable phase shift is exactly how each tower gets its assigned timing). The coils are big edge-wound copper ribbon with taps or rollers; the capacitor is a vacuum or mica unit rated for thousands of volts. The static drain choke looks like one more coil but carries no matching duty; it's a high-reactance path that quietly bleeds rain static and induced charge off the hot tower so it can't build up a wallop. Values shown are a real solution for a quarter-wave tower at 1000 kHz; a real installation lands near these numbers, then the engineer trims taps until the common-point meter reads exactly right.

Drawing 2.4-2

THE AUSTIN RING TRANSFORMER: LIGHTING A TOWER YOU CAN'T TOUCH beacon side light Whole tower is HOT thousands of volts of RF: no wire may connect it to grounded equipment 240 V lighting feed the rings, up close AIR GAP the rings never touch Secondary ring: HOT hangs from the tower side, feeds the beacon & side lights Primary ring: grounded bolted to its pedestal, fed 240 V from the lighting circuit 60 Hz magnetic flux threads both rings (power crosses the gap, RF doesn't)

Two chain links that never touch. A series-fed tower carries thousands of volts of RF, so you cannot simply run a lighting wire from the grounded world up the steel; it would short the antenna to ground. The Austin transformer solves it like a magic trick: two toroidal rings interlinked like links of a chain, each wound as one half of a transformer. The grounded primary ring sits on its pedestal and takes ordinary 240 V lighting power; the hot secondary ring hangs from a bracket on the tower, inches away across an air gap. The 60 Hz magnetic flux happily jumps the gap and powers the beacon and side lights, while the same gap presents almost no capacitance, so the tower's RF stays on the tower. Invented by Arthur O. Austin in the 1920s, the design is so right that towers built last year still use rings that look just like the originals. You can spot one at almost any series-fed AM site: look for the interlocked hoops beside the base insulator.

Drawing 2.4-3

Interactive · Antennas

The Capacity Hat: making a tower taller without adding steel

On a bare tower the current tapers to nothing at the top, so the last third of the steel barely radiates. A hat puts current up there. But watch what it actually buys you: the answer is not the one most people expect, and it depends entirely on what's buried under the tower.

Current on the steel

Drawing 2.4-4

Where the transmitter's watts actually go

Drawing 2.4-5

 Bare towerWith the hatChange
Effective height
Radiation resistance
Efficiency
Field at 1 km per kW
Base reactance

Table 2.4-6

The surprise, and why it's real. Radiation resistance rises as the square of effective height; this calculator confirms it to four figures. So for a lossless tower, hatting it raises heff and Rrad together and the radiated field barely moves. The hat is not a gain antenna. What it actually does is raise Rrad until the ground system's loss resistance stops mattering: on 120 radials it's worth a fraction of a dB, on a compromised ground system it's worth several. It also walks the base toward resonance, which drops the base voltage and widens the bandwidth the sidebands have to live in. Exact: the current distribution I(z) = sin(G+T−z), the radiation integral over the monopole and its image, Rrad, heff and the field: the engine reproduces a bare quarter-wave at 36.56 Ω and 313.8 mV/m per kW at 1 km, matching the other explorers on this site, and its radiation integral agrees with the independent induced-EMF formulation to within 0.000% across 30-170°. Approximate: base reactance uses a transmission-line model X = −Z₀·cot(G+T) with Z₀ from tower slenderness, and the radial-count-to-ohms mapping is indicative only; real ground loss depends on soil, radial length and burial, and is measured, not calculated. Hat radiation is assumed to cancel in the horizontal plane, which holds for a symmetric hat. 73 de N6JET

Interactive · Tuning house

The ATU Bench: design a T-network

The worked example on this page matches a 50 Ω line to a 36 + j21 Ω tower at 1000 kHz. Here it is as a bench: change the frequency, the tower, or the phase shift, and watch the coils and capacitor resize. The phase shift is the knob that matters: in a directional array, that's how each tower gets its assigned timing.

Tower base resistance (Ω)
Tower base reactance (Ω, signed)

Inside the doghouse

Drawing 2.4-7

LegRequired reactanceComponentValue
Series: line side
Shunt: to ground
Series: tower side

Table 2.4-8

The arithmetic. An image-matched T‑network transforming R1 to R2 with phase shift θ needs shunt Xb = −√(R₁R₂) / sin θ, series Xa = (√(R₁R₂) − R₁·cos θ) / sin θ on the line side, and Xc = (√(R₁R₂) − R₂·cos θ) / sin θ on the tower side, from which the tower's own reactance is subtracted, because the tower supplies part of the output leg for free. At θ = 90° every leg collapses to |X| = √(R₁R₂), which is the shortcut used in the worked example. Set it to 1000 kHz, 90°, 36 + j21 and this bench returns 6.75 µH, 3,751 pF and 3.41 µH; the page says 6.75, 3,750 and 3.4. Base impedance in "compute from height" mode comes from the thin-wire monopole integrals and is an idealisation: a real tower's cross-section, guys, base region and ground system move it, which is why an engineer measures the base Z rather than trusting a formula. Positive phase = lag (low-pass, series coils); negative = lead (high-pass, series capacitors). Component ratings, common-point metering and the static drain choke are not part of the match. 73 de N6JET

© 2026 Chris, N6JET. Free to read and share for non-commercial use with credit (Creative Commons BY-NC 4.0). Not for sale.

More in The Tower and Antenna

All AM radio stories by subject · AM radio interactive tools and simulators · AM Trivia Challenge quiz