Line Arrays

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
What do people think of line arrays? They seem to be a pretty good and simple way of using many relatively inexpensive drivers to build a decent speaker. I haven't noticed much information on building them, however. Is it as simple as just stacking multiple drivers on top of each other?

Also, about the wiring. Usually, multiple drivers are wired in some series/parallel combination that puts them near the 4-8 ohm range, but there's been at least one disucssion hereabouts regarding low (insert more o's in there...) impedence loads. I think someone wanted to parallel 32 driver or something? What about speakers with very high impedences? Like 50-100 ohms? The amplifier would need a bigger voltage swing, but this is better for class-AB amps, anyway. I would imagine that most amplifiers drive larger loads more stably and with less distortion.

Regards, Won
Line source speakers are capable of excellent sound reproduction. How well an array of smaller speakers works depends on the frequency you're trying to reproduce. At low frequencies relative to the spacing of the individual drivers, multiple drivers work well. Unfortunately, at higher frequencies, the wavefront no longer combines; it breaks into lobes and nulls.
The Loooooow Impedance thread interested me, as I like ribbon drivers, which are (with the exception of the Magneplanar ribbon) so low in impedance that they are generally driven with transformers. Since I use the Magneplanars, it's not strictly necessary, but I wanted to fiddle with the idea anyway. If nothing else, the very extremity of the concept attracts me. At the moment, however, I'm tied up trying to get a Mini-Aleph finished, and after that, I'd like to get to, the "Clamor" circuit. You know...<i>that</i> one.
On the other end of the spectrum, higher impedance loads are indeed easier to drive. That's an interesting idea also, but not quite interesting enough to force its way into my queue of things I want to play with. As a passing hint, note that tubes just love to swing voltage...something like a 6AS7 or 6C33 in the right circuit would be happy as a clam if you were to give it 30-50 ohms to play with.

User YA posted his line array spreadsheet on the Madisound board a few weeks back, it's quite instructive, his page is at

(look under Share).

My understanding is that an array needs to be at least 70% of the floor-ceiling height to be considered a true line source. I think the height is needed so that, in theory anyway, the floor and ceiling reflections cause the array to mimic an "infinite" line source.

That's all a bit much for me, I'll be experimenting with a "short array" ie four drivers on a side. Suboptimal but at least it's somewhat affordable! The spreadsheet above indicates that four drivers is not too bad in terms of comb filtering, any more and you have to go the full length. GR Research has the new BG Neo8 ribbons for $60 (I think) apiece. He also has a line array kit, in case you were interested in that...

Well, sorry for the long-winded post, hope some of this is useful!

A line source has to be longer/taller than, I think, a half-wavelength at the lowest frequency you intend to reproduce. This isn't a problem at 20kHz, but gets trickier at lower frequencies. Reflections come in handy.
My subs and mid/tweeter arrays are a whisker over 7' tall (eight foot ceiling except for where it shelves down to 7'4" at the front of the room). The woofer panels are 6', but are more properly considered planar speakers, so I can't claim to be line source from top to bottom. One of these days, I'm going to try a thing or two to see if I can sneak past that limitation, but it'll have to wait.
The Bohlender-Graebener RD-75 makes a very nice line source mid range unit, but don't try to run it all the way out to 20kHz as it gets quite soft sounding on the top. It's good down to about 200Hz on the bottom.

speaking of arrays...

Won, there is a friend of mine who is currently designing a DIY line array pair. His name is Rick Craig, and he is a respected member of the DIY community (and has a good reputation on the Madisound list). I am told that he is near completion, and he recently posted a message to that list here saying so. If you can wait a short while, he will release his cabinet and crossover design along with his bill of materials for anyone to build.

One of the primary factors of this project was the cost. I believe Rick says you can build them for $700/pr (excl. cabinets).

Grey : Rick is using ribbons in his design. If you are a fan then you might want to take a serious look at this. :cool:
ucla88, I don't think Danny has anything on his web page. He posted about it on, but that site is down now :-(. I think parts express was going to sell it as well. The data sheet is on They have a peak at 12 kHz, according to Danny this goes away if you have eight of them, I am not sure however if some eq is needed to bring the top end up.

I am planning to use four of the Neo8s per side instead of a full array ($$!). I will also try four midranges I got from Madisound next to them, not sure yet if I can make it work as a two-way. Anyway I haven't got the ribbons yet, I need to sell off some stuff on EBay first (sort of like the rotating equipment merry-go-round :-()

Won, OTL amps do like higher impedance loads, up to a point that is when output is limited by supply voltage rather than current delivery. But a 16 or 32 ohm load is certainly more favorable to an OTL. There are commercial OTLs by atmasphere and transcendentsound for quite a while, and there are a whole slew of newer ones that also cost a pretty penny (Joule, Graaf, I forget who else). There are also many schematics on the net, if you are interested let me know and I'll try and list as many as I can. I plan to wire my mid and ribbon arrays as 16 ohm loads and build low power OTLs for them.

The Atma-Sphere amps are quite happy with an 8 ohm load (4 ohm too, for that matter). But, yes, they are also good for higher loads. I haven't paid as much attention to the other models. The old Futterman layout is always a possibility, too.
Diy OTLs are not difficult, and once you're free of the output transformer, amazing things are possible (Bandwidth to a quarter megaHertz? Is that good enough for you? Man, you can talk to bats with these things, and they're stable beyond anything solid state people can even dream of...). I've got two different designs I've been tinkering with...but you know the line...time, I need more hours in a day.

Man, this thread has all my favorite hot-button words: line array, high impedence, low impedence, OTL amps...

I'm just drawn to that wacky stuff. Maybe it's because the rest of you wise ones already have the mainstream honed so fine, you make it look easy.

Yes, my interest ranges from 1/2 ohm to multi-Kohm speaker loads--with not much in between. (I'm the guy who wants to wire a line array of 32 8-ohm drivers in parallel, or wind one 2Kohm full-range driver.)

Anyhow, I currently listen to a pair of line arrays I built--12 4-ohm 4-inch wide-range drivers per side in 80L bass reflex boxes tuned to about 48Hz. They are very efficient and capable of *BIG* sound at very low excursion. The bass energy was a surprise to me--Not all that deep, but extremely solid and coherent. Many hands make light work. Even during a 50-watt blast (and at 96+dB/W 50 watts is a blast), each driver sees only a little more than 4 watts. Normal listening pushes them with only a fraction of a watt each, so they operate in a very linear range of motion.

My original idea was to try a bessel array (mathematically approximates point-source dispersion), but I read that the wavefront didn't unify until it had propogated at least 3x the array hight. So I decided to try a tapered array, ala D'Appolito.

In my case, I wired the center 4 drivers in series--these are the acoustic center. I wired four more drivers and two parallel pairs in series. Finally, I wired these two series lines in parallel. The second four go two above and two below the first four, and the parallel pairs are on the ends, two on top and two on the bottom. The net effect is power delivery that is concentrated in the center four drivers of each line and tapers off to the ends. In my case, the nominal impedence for each line is 8.88 ohms. Crossover is 1st order at 6KHz to a cheap dome tweet.

By wiring a tapered array, I sacrificed some overall efficiency for (I hope) better imaging. Wired normally, each line would have given me about 99dB/W.

(An aside: purchasing the drivers on an internet auction and roughing the cabinets together myself, I spent a little less than $100 on this project:))

One thing I do wonder about is comb filtering in line arrays. (Anybody feel free to correct me if I'm mistaken):

My understanding is that pressure waves from each driver in a line array interact with pressure waves from every other driver, independant of what is between them. (Waves pass through each other, summing to anything from zero to 2 depending on angles of phase and intersection at all points in space.)

Therefore, the top and bottom drivers of a line array, being far apart, will interact to create comb filtering beginning at quite low frequency. The fact that other drivers are firing between the extremes (and setting up their own different comb filter interactions) would tend to mask the response ripples set up by any two drivers, but those ripples are still there. In a typical line array of 10-20 drivers, the interactive combinations are very numerous and I imagine that the comb filter interactions are extremely complex. Do all these ripples of slightly different phase combine to form flat response? I don't know.

I get a kick out of my line arrays, but I still wonder if I'm truly ascending to higher levels of Fi.

I've got these two cases of 32 full-range 4" drivers each, just burning a hole in my basement floor, so I bull-headedly continue to dream of ways to use them--all at once.

What's currently on my brain is a "matrix" array, for lack of a better word. I'm picturing a 32"x16" open baffle with a 4x4 (=16) driver matrix in the middle. This will be the midrange. Since all the driver centers will be close to their fellows, comb filtering shouldn't begin till higher in the band. Above and below the 16-driver matrix, I'll put 16 more drivers (8 above and 8 below) and roll them off around 150 to fill in the lower end.

This design would probably give up some horizontal dispersion and add some vertical dispersion compared to a line array (though open baffle cancellation would lessen room reflections).

Anyway, whatta y'all think?

I believe Wayne J over at SpeakerBuilder tried something like this using the "wonder" drivers that were a Parts Express special, with a hefty matrix-style array. He angled the rows of the array slightly outwards each time, giving a slight curvature. However, he gave up after realizing that the crossover needed to tame the problems would be far too complex.
So what are different ways to wire a line array? It sounds like you made it such that the center drivers recieve the most power (I didn't get that much from your description Bill...I should try rereading it). What's the differences between distributing power equally or putting more power towards the ends?

Also, I'm also considering building a "short" line array of 6x6.5" midbass woofers and 2 dome tweeters (MMMTTMMM) that would stand around 4' tall. Is this a sufficient number of drivers/height to be considered a line source?

Thanks, Won

(This is kinda long)

The cool deal with a perfect line source (infinite length, single transducer) is that it launchs a hemicylindrical wavefront. This pressure wave is different from a point-source hemispherical wavefront in that its width spreads as it travels outward, but its height remains the same (theoretically infinite). A point-source wavefront spreads in width and hight and its pressure decays at 6dB per doubling of distance, but a perfect line source wavefront, spreading in only one dimension (width), decays at only 3dB per doubling of distance.

This effect is a definite bonus if you're trying to illuminate long distances as evenly as possible, such as a crowd at a large concert, ergo the popularity of large line array PA systems.

There is no infinitely tall loudspeaker, so there is no such thing as a perfect self-contained line source. However, if you physically limit the height dimension with a floor and ceiling and illuminate that space with a floor-to-ceiling line source, you still end up with a fine hemicylindrical wavefront and all the joy that it brings.

Speakers that are not infinitely tall or loaded by a ceiling also launch a hemicylindrical wavefront, but that wave will begin spreading vertically. Over distance, it will become more hemispherical and tend toward the corresponding 6dB decay rate.

I don't remember exactly, but I think the wavefront from a finite line souce in free space loses much of its line source characteristic after traveling 2x the line height. So if you want to hear line source effect from 4' tall array in free space, you'll probably have to keep your head no more than 4' from the middle of it.

Incidentally, line arrays (multiple drivers) in free space only approximate a line source at wavelengths longer than total array length. At higher frequencies, the beams begin to decouple and develop interference effects--lobing & comb filtering. (See my previous post about line array comb filtering.) Also, in your MMMTTMMM situation, the Ms will behave more like a line source than the Ts will, so they will have different decay rates (highs will drop off faster as you move away).

All this is to say that, strictly speaking, a line source isn't a line source unless A) it is loaded by a floor and ceiling or B) you sit within its shadow.

And none of this is to say that a 4' tall MMMTTMMM quasi-not-really line source couldn't sound glorious:)

As I've read it, the benefit of tapering the power delivery is rooted in psychoacoustics: Listening to a vertical array of speakers, your brain will tell you that all sound is coming from A) the loudest speaker or B) the speaker nearest your ear. So, in a set of non-tapered arrays, the center of the stereo image will move vertically as the level of your ears move.

I chose to fix the vertical reference point by making the center four drivers louder than the rest. It makes my array behave less like a line source and throws away some efficiency, but I wonder if it might improve the comb filtering situation since the drivers farthest apart are also the quietest.

I know, I know: you're asking what the point of a diluted line array is. I dunno. Just playing around and having fun. I ran the wiring externally so I could remove the taper if I got the notion.

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.