Line Array question

[aceinc]

When does a group of drivers in a vertical line become a line array?

Is there a minimum heght? 6"? 12"? 24"?

Paul

[karelg]

There is a very good paper on line arrays by James Griffin:
http://www.audiodiycentral.com/resource/pdf/nflawp.pdf

[Few]

Do look at Griffin's paper for a more detailed description, but for a very quick and dirty response I'll add that it's the source's dimensions compared to the wavelengths of sound being reproduced that determine whether the radiation pattern approximates an ideal line source. The source has to be significantly taller and narrower than the wavelength of sound being produced in order to mimic the expanding cylinder-shaped wavefront associated with a line source.

Also, if the listener is very far from the source compared to the source's height, even a tall narrow source ultimately starts to be perceived as a point source. For example, a 6" tall tweeter ribbon will have more restricted vertical directivity than a 3/4" diameter dome. But, if you listen to it from 20 feet away, the loudness will start dropping off as 1/distance squared (like a point source) instead of dropping off as 1/distance (like a line source).

Few

[aceinc]

Thanks for your responses.

The reason for my question is that I am deciding on the layout of my next set of speakers. I have twelve drivers per cabinet, four each 5", 3", and 5/8" dome tweeters. Perhaps someone can give me guidance on how best to utilize these drivers.

Some of my ideas include (Not to scale);

[Few]

This isn't a complete response, but I'd recommend using only one tweeter per speaker if the choices are four tweeters or fewer. You'd need far more than four tweeters---with very small flanges so they can be closely spaced---to make an effective line source for the highest frequencies. A truncated array of tweeters of 2, 3, or 4 tweeters is likely to cause more trouble than its worth.

I'd also recommend arranging the other drivers vertically rather than side by side. Rather than focusing on a line array you might consider symmetric arrays with a central tweeter and larger drivers arranged above and below it. To some extent, it depends on how sophisticated a crossover you want (or are able) to create. If you augment the speakers you've descrived so far with some bass drivers, you'll end up with some variation of a four-way speaker: not a trivial undertaking. A simpler design done well might yield more satisfying results. On the other hand if you're a master crossover designer, all sorts of combos can be considered.

Few

[bjorno]

Hi Paul,

Take a look at your original picture I modified and played with and the 4 drivers in column polar plots that may be of use when deciding driver layouts and eventually for the crossover considerations that you might have.
See pictures 1(8)-7(8)

When does a group of drivers in a vertical line become a line

My array definition and other suggestions:

A line array in my homebrew definition always uses a sufficient number of drivers (n>=12) where the peak directivity coincide within a dB of a full wavelength = the c-c distance between the drivers, all other arrays using less number of drivers (short arrays) should in my opinion be referred as to be column speakers (n<12).

Jim Griffins recommendations for Nearfield Line arrays and use of lambda and lambda/2 criteria’s is quite in order due to the fact that the number of drivers used is >= 12 when using M (<= 5”) and T drivers, the former covering up to 70 % of the room height.

Quote Jim Griffin:

‘2. Near field. Urban, et al [1] derives a more restrictive criterion of no more than a half wavelength separation between drivers at their highest operating frequency…. This analysis is based upon their desire to place any far field dips (nulls) in the angle off axis response of the array beyond p/2 (90 degrees). This assures that secondary (off-axis) lobes in the sound field are greater than 12 dB down from the on-axis response (main lobe)’…

I find this very important to consider and even good if exceeded when designing line arrays (M drivers) and when fewer than 12 drivers is used in a shorter column speaker, sharpening of both the lambda and the lambda/2 criteria improves in my opinion the spatial stage impression substantially.

The c-c distance between the used drivers (M) is a measure of the line array usable upper corner frequency and this distance is usually described in terms of a whole or a fraction of a wavelength.

Another measure that could be used as a figure of highest usable frequency could be based of directivity loss above the peak directivity and within 1.5 x lambda.

A 12-driver array has a drop of about 8 dB to the mean directivity value between lambda and 1.5 x lambda.

For a column of 4 drivers the directivity drop is lower, about 2.7 dB as can be seen in picture 1(8).

When using criteria like lambda or lambda/2 for a line array this number must be accurate enough to point out where the directivity suddenly drops without causing errors when calculating the crossover attenuation at that point.

I.e. if lamda/2 is found to be sufficient used with a 12-dB/octave filter to suppress the ripple coming up at lambda then the number of drivers must exceed 12 where the directivity peaks close enough to lambda = 344/(c-c distance).

If n = 12 then the peak occur at 0.97x 344/(c-c distance) and for this case when about 0.97 can be rounded off to equal 1.

For all other lower numbers of drivers the Lambda value should to be lowered with a factor: about 0.92 x for 10 drivers, 0.9 for 8,0.86 for 7, 0.84 for 6 and 0.79 for 4 drivers.

If the M driver crossover is in the 3 kHz regions, where the ear amplitude sensitivity peaks*, the hearing can be said to be equipped with amplitude zooming (higher resolution) by aid of the ear channel dimensioning.

(If the T driver has high Q HP transfer function and fs is too close to the crossover point or where a high order HP crossover is used, crossover driver combination may cause ringing transient ripple that easily can be heard).

*The ear channel provides the eardrum with about 10 times the pressure in a window of about 2- 5.5kHz when compared to the outside folds.

Amplitude errors should be avoided here and for an array the upper frequency range smoothness should in my opinion reach up to 8-9 kHz in order to suppress ambiguous hearing.
See Picture 8(8).

The fact that an 40 “ array at 10’ listening distance occupies almost the vertical maximum MAA resolution azimuth if centred at the ear height, any blooming or side loobing enlarges the soundstage and is easily detected if present and labelled typical array footprint sound.

Luckily phase has lost most of its dominating impact for hearing above 1.5 kHz but instead its where the amplitude plays a more important roll and if crossing over in the vicinity of 3 kHz is chosen for the M driver and if a frequency about 3/4x lambda or higher up is used, then a very steep filter (24 – 48 dB/ octave) is necessary to push down the ripple at lambda and the secondary loobing to a minimum.

An unspecified (LP higher order) type of filter is adequate and the only thing to watch out besides for a mandatory flat FR is the horizontal loobing that is a resulting function of filter inter driver phase and the c-c distance.

When placing the crossover in the 1-2kHz(W-M) regions (See also picture 8(8)) where phase and amplitude localisation impact is about equal worth for the hearing, an interesting observation can be made:

The T line (normally always close to the M line) can be moved a substantial distance to form a lower azimuth angle (46) than the normal of 60 degrees where otherwise vertically aligned speakers are placed.

This can be done with success only if low order filters (<= 2nd) is used for both M/T drivers and the T drivers have fs about 500-600Hz together with a usable bandwidth reaching down below 1.5 kHz.

In the case of using higher crossover point like between 2-3 kHz its better to place W drivers at 30 degrees and M/T close at 46 degrees leaving a gap between W and M/T.

See picture at: http://www.diyaudio.com/forums/attac...amp=1179074994

This layout improves the central phantom sharpness i.e. MAA (Minimum Audible Angle) coincides or is minimized due to the fact that the ear brain combination locates combination cues from 2 speaker phantom listening differently depending of spectral bandwidth and speaker set-up azimuth.

Read this short but very informative resume of sound localisation at:

http://www.physicstoday.org/pt/vol-5.../locsound.html


b

1(8)

[bjorno]

2(8)

[bjorno]

3(8)

[bjorno]

4(8)

[bjorno]

5(8)