AM Directional Antenna Engineering

A Tale of Two Arrays

How the directional pattern of an AM station is decided — worked through two real, contrasting stations: KZAC 560 (San Francisco), a textbook two-tower cardioid, and KNTH 1070 (Houston), a switched eleven/nine-tower array that reconfigures itself at sunset.

The one idea to hold onto: an AM directional pattern is never picked from a catalog of nice shapes. It is derived as the solution to a constraint problem — protect every other station on the channel from interference, while covering the community you are licensed to serve. Everything below is that trade-off playing out under two very different sets of constraints.

The decision framework

Before the two cases, the four inputs that determine every AM pattern:

Who must you protect? AM channels are shared. The rules cap how much signal you may radiate toward every other co-channel and adjacent-channel station's protected contour. Those caps, plotted by compass bearing, dictate where the pattern needs nulls.
Day or night? By day, AM travels by ground wave only — a local, geometric problem. At night the ionosphere returns sky wave hundreds of miles, so protection tightens dramatically and often forces a separate, more directional night pattern.
What is the site worth? Ground conductivity (salt water is superb, dry rock is poor) and the transmitter's position relative to both the audience and the protected stations shape what is easy or hard to achieve.
How many nulls do you need? Each independent, deep null in a specific direction costs a degree of freedom — and degrees of freedom come from towers. One direction to null → two or three towers. Many directions → a large array.

That last point is the whole difference between our two stations.

KZAC560 kHz Class BMinimum case

San Francisco, CA · transmitter at Islais Creek on San Francisco Bay · FCC Facility ID 34472

Day power

5 kW

Night power

5 kW

Day pattern

Non-dir.

Night towers

Night RMS

681 mV/m

Pattern

Cardioid

Night pattern (schematic, geographically oriented). The single deep null points ENE toward Denver; every protected 560 station falls in that suppressed north-through-east arc, while the main lobe covers the city and coast to the west.

Radiated pattern Primary protectee Other 560 stations

Why one cardioid is enough

560 is a regional channel shared by a handful of full-power stations — and from a San Francisco transmitter, all of them sit in the same slice of sky. Every protected station lies between due north and due east:

Protected station	Bearing	Distance
KPQ Wenatchee WA	8° N	1,090 km
KMON Great Falls MT	36° NE	1,410 km
WIND Chicago IL	70° ENE	2,980 km
KLZ Denver CO	76° ENE	1,520 km
KWTO Springfield MO	82° E	2,560 km

Because the entire protection problem lives in one quadrant, a single null solves it — and a single null is exactly what a two-tower array makes. Feed the two towers the right relative phase and you get a cardioid: one broad lobe, one deep notch on the opposite side. Aim the notch ENE at the Denver cluster and the job is done.

Two more things fall out for free. By day, ground wave can't reach Denver at 5 kW, so KZAC runs non-directional — the array only matters after dark, when sky wave switches on. And the Bay-edge site gives excellent ground conductivity for the lobe thrown across the water and the city.

KNTH1070 kHz Class BComplex case

Houston, TX · transmitter at 29°59′33.8″N 95°28′23.8″W (NW Harris County) · FCC Facility ID 61174 · licensed to Salem / South Texas Broadcasting, engineered by Carl T. Jones Corp.

Day power

10 kW

Night power

5 kW

Day towers

Night towers

Night RMS

909.3 mV/m

Pattern

DA-2 switched

Night pattern reproduced from KNTH's published FCC theoretical pattern, geographically oriented. A single main lobe points due south (slightly east) — straight at Houston and the Gulf — while KNX (WNW) and the whole northern protected cluster fall in the nulled region. Red is true scale; blue is the same pattern at 10× zoom (the FCC's own convention), showing the small side lobes toward SE and SW. Bearings are true.

Radiated pattern KNX (Class A) Other 1070 stations 10× zoom (FCC style)

Why it takes nine (and eleven) towers

1070 is a clear channel owned by a 50,000-watt Class A giant, KNX Los Angeles. As a secondary Class B, KNTH is allowed on the channel only if it rigorously protects KNX and a scatter of other stations spread right around the dial:

Protected station	Bearing	Distance
KNX Los Angeles (A, 50 kW)	288° WNW	2,190 km
KFTI Wichita KS	349° N	870 km
KHMO / WTSO (MO·WI)	18° NNE	1,140 km
WFNI Indianapolis IN	35° NE	1,380 km
WDIA Memphis TN	40° NE	770 km
WAPI Birmingham AL	62° ENE	910 km
KOPY Alice TX	226° SW	355 km

Now the nulls are demanded in many directions at once — WNW, N, NNE, NE, ENE and SW. You cannot sculpt that many independent deep notches with a handful of towers: each one costs a degree of freedom, so the array grows to nine elements at night to place them all while still filling a usable lobe over greater Houston.

And because ground-wave-only daytime and sky-wave-active nighttime are two different physics problems, KNTH licenses two separate arrays — 11 towers at 10 kW by day, 9 towers at 5 kW by night — and physically switches which towers are driven, and how they're phased, at sunrise and sunset. That is the defining mark of a genuinely complex AM facility.

The licensed night array — the real 9 towers

The actual per-tower parameters from KNTH's FCC license (application BMML-20170310ABA). The pattern above is computed directly from these numbers — nothing invented:

Twr	Field ratio	Phase	Spacing	Orientation
1	0.258	100.8°	0°	0°
2	0.511	1.8°	90°	160°
3	0.258	−97.2°	180°	160°
4	0.505	99°	210°	230°
5 (ref)	1.000	0°	90°	160°
6	0.505	−99°	90°	160°
7	0.258	97.3°	420°	230°
8	0.511	−1.7°	90°	160°
9	0.258	−100.7°	90°	160°

All nine towers are 196° tall (152.54 m). Spacing and orientation are electrical degrees, chained tower-to-tower. Theoretical RMS 909.3 mV/m; the licensed pattern also carries 13 augmentations — small localized bumps added to the standard pattern — which are not modeled in the plot above.

Side by side

Parameter	KZAC 560	KNTH 1070
Channel type	Regional	Clear (secondary)
Must protect	One quadrant (N–E)	Most of the compass
Marquee protectee	KLZ Denver (5 kW B)	KNX Los Angeles (50 kW A)
Day power / pattern	5 kW · non-directional	10 kW · 11 towers
Night power / pattern	5 kW · 2 towers	5 kW · 9 towers
Day vs night arrays	Different (ND → DA)	Different & switched (DA-2)
Pattern shape	Simple cardioid	Multi-lobe, multi-null
Independent nulls	~1	Six-plus

The principle underneath

The two stations are the same equation with different numbers. Count the directions you must suppress, and that sets the size of the array. KZAC's protection problem collapses into a single quadrant, so one null — one cardioid, two towers — disposes of it, and the pattern only needs to exist at night. KNTH's protection problem is smeared across the whole horizon and headlined by a 50 kW clear-channel station, so it needs many simultaneous nulls, which means many towers, plus a second full array because day and night are different problems.

Neither pattern was chosen for elegance. Each is simply the smallest, cheapest array that keeps this station's signal off everyone else's while still covering home — and that single constraint, run through two different channels, is enough to produce a tidy cardioid on one coast and a switched eleven-tower machine on the Gulf.

On the diagrams: Station facts — power, tower counts, RMS field, class, coordinates, facility IDs — are from FCC records; bearings and distances are computed great-circle values. KNTH's night pattern is traced from its FCC-published theoretical pattern (the red/blue polar plot on the license), so the lobe shape matches the licensed pattern faithfully. The tower table below it is the real drive data; computing the pattern from that table alone requires the exact tower-to-tower layout, which the license's chained spacing/orientation columns didn't fully resolve — so the traced plot is the accurate one here. KZAC's two-tower cardioid is still a schematic — pulling its FCC record would make it exact too.

Sources: FCC AM station records via radiostation.info (KZAC) and radiostation.info (KNTH) · KSFO/KZAC history · KNTH · 560 kHz allocations · 1070 kHz allocations · KNX Los Angeles · FCC license application BMML-20170310ABA, Carl T. Jones Corp. (KNTH night array)