> ## Content Index
> Fetch the complete content index at: https://abovethefader.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The J isn't a style choice: how a line array actually works
- URL: https://abovethefader.com/the-j-isnt-a-style-choice-how-a-line-array-actually-works/
- Published: 2026-09-23T19:09:23.000Z
- Updated: 2026-09-23T19:09:22.000Z
- Description: Six dB per doubling, half a wavelength, and eighty percent of a ribbon. Why the top boxes are flat, the bottom ones are fanned, and the prediction plot was never the room.
- Author: Tal Kocen
- Tags: PA & System Design, The Woodshed

Walk into any arena load-in and you will see the same shape. A dozen boxes in the air, the top few hung almost flat against each other, the bottom few fanned out like a hand of cards. It reads as an aesthetic decision, or as whatever the rigger felt like that morning. It is neither. That curve is the whole argument, and it falls out of a piece of physics the industry took about forty years to accept.\\n\\n## The problem with one big speaker\\n\\nStart with a single box. Point source, radiating roughly spherically. Double your distance from it and the same energy spreads over four times the area, so you lose 6 dB. That is the inverse square law and it is not negotiable.\\n\\nSix dB per doubling is brutal over arena distances. Ten metres from the box at 100 dB puts you at 94 dB at twenty metres, 88 at forty, 82 at eighty. To get the back of the room loud enough you have to cook the front row. Anyone who has mixed on a wall of point-source boxes knows that trade.\\n\\nThe obvious fix is more boxes. Point them at different parts of the room, splay them wide, cover everything. That works, sort of, until you notice the seats where two boxes overlap sound strange. Comb filtering. Two arrivals of the same signal a few milliseconds apart, summing and cancelling their way up the frequency range. You have traded a level problem for a coherence problem.\\n\\n## Coupling, and the condition for it\\n\\nHere is the thing that makes a line array a line array. Put two sources very close together relative to the wavelength they are producing and they stop behaving like two sources. They behave like one bigger source. The wavefronts leave in phase, they sum coherently, and you get 6 dB more output instead of the mush two boxes usually make when they argue.\\n\\nVery close relative to the wavelength is the whole game, and it is why line arrays are hard.\\n\\nTake 100 Hz. Wavelength about 3.4 metres. Two boxes stacked half a metre apart are, at that frequency, practically on top of each other. They couple without anyone doing anything clever. Low end is easy.\\n\\nTake 10 kHz. Wavelength about 34 millimetres. Those same two boxes are now fifteen wavelengths apart. At that spacing they do not couple. They interfere, and the result is a lobed, ragged mess that changes every time you move your head.\\n\\nSo a stack of ordinary boxes is a line source in the low end and a pile of arguing tweeters up top. Which is exactly what large arrays sounded like before anyone solved it.\\n\\n## Making the high end behave\\n\\nChristian Heil and Marcel Urban published the criteria in the early nineties and the commercial result was the V-DOSC. The requirement runs roughly like this. For a column of sources to behave as one continuous line at a given frequency, the radiating elements have to sit within about half a wavelength of each other, and the radiating surface has to be substantially continuous. The figure usually quoted is at least 80 percent of the line actually radiating, with no meaningful gaps.\\n\\nYou cannot do that with cone drivers at 10 kHz. Half a wavelength is 17 millimetres. No compression driver and horn is 17 millimetres tall.\\n\\nSo you cheat, and the cheat is the interesting part. Every line array on the market has some device sitting between the compression driver and the air whose only job is to take the output of a small circular source and turn it into a tall, thin, flat wavefront that leaves the box as a vertical slot. Manufacturers do it differently and they all have their own name for it. The goal is identical: make the high frequency section behave, geometrically, like a continuous ribbon running the height of the box. Stack the boxes and the ribbons join into one tall ribbon. That is the trick. Everything else follows from it.\\n\\n## What you get for the trouble\\n\\nA true line source does not radiate spherically. It radiates cylindrically, expanding sideways but not vertically. The energy spreads over the surface of a cylinder rather than a sphere, and the area of that cylinder grows linearly with distance instead of with the square of it.\\n\\nWhich means 3 dB per doubling of distance, not 6.\\n\\nThat is the payoff and it is enormous. Over four doublings you have saved 12 dB. That is the difference between a front row that hurts and a front row that does not, at the same level in the back.\\n\\n## Why the array is curved\\n\\nNow the J.\\n\\nA perfectly straight array throws a flat wavefront straight out. Fine for the far seats, useless for the ones underneath it. So you curve the bottom.\\n\\nThe top boxes stay tight, sometimes at zero or half a degree of splay, because they are doing the long throw and you want maximum coupling over the greatest distance. The lower boxes get progressively more splay, three degrees, five, eight, because they are covering seats that are close and steeply below. Wider angle, more vertical coverage, less throw. You are spending coupling to buy coverage exactly where you need it.\\n\\nThat is a J array, and the shape is a direct readout of the room. A deep bowl with a steep upper tier gets a different curve than a flat field. When you see a rig hung in a smooth constant curve instead of a J, someone decided that room wanted even coverage more than it wanted distance.\\n\\n## Where the theory stops working\\n\\nTwo honest caveats, because the cylindrical-wave story gets oversold.\\n\\nAn array only behaves as a line source out to a certain distance, and past that it goes back to acting like a point source. The transition depends on the length of the array and the frequency, scaling with the square of the array length divided by the wavelength. Long array and high frequency, the line behaviour holds a long way out. Short array and low frequency, it reverts quickly. This is why a six-box hang does not get you arena performance no matter how carefully you splay it, and why the low end of any array is back to 6 dB per doubling long before the top is.\\n\\nThe second caveat is air. It absorbs high frequencies over distance, more or less depending on temperature and humidity, so the back of the room is always duller than the front. System techs deal with this by shading the array, giving the long-throw top boxes more high end than the down-fill boxes so that what arrives at the far seats has roughly the tonal balance of what arrives at the near ones. That is what the per-box gain and EQ offsets in the prediction software are for. Not fine-tuning. Compensation for the atmosphere.\\n\\n## What this means for the gig\\n\\nIf you mix on line arrays and never touch the rigging, the useful part is this. The array is a single instrument rather than a collection of boxes. Its vertical coverage was decided when it went in the air, and no amount of EQ at the desk fixes a splay angle. If the top of the room sounds thin and the floor sounds thick, the problem is upstream of you and the conversation is with the system tech.\\n\\nAnd when someone hands you a prediction plot showing beautiful even coverage, remember what it is a model of. It assumes the boxes are where the drawing says they are, the room is empty, and the air is whatever the software defaulted to. The moment the room fills with bodies the high end changes. The plot was never the room. It was an argument about the room.\\n\\n— *Above the Fader*