Line Array speaker Design?????

Discussion in 'Archived Threads 2001-2004' started by rich r, May 8, 2002.

  1. rich r

    rich r Stunt Coordinator

    Joined:
    Jul 29, 2001
    Messages:
    56
    Likes Received:
    0
    Trophy Points:
    0
    Hey all,

    I am interested in designing and building a "line Array" speaker using 6+ drivers and a tweeter. Does anyone have and pictures, designs or thoughts that they would like to share, to help me fuel this design??

    Is there any precautions that should be considered when designing a line array??

    I would like to overall effiecency to as high as possible (93+)and also have a crossover design using as few components as possible, also while still covering the broadest bandwith possible.

    thanks for anyy input in advance?
     
  2. Line arrays are very tricky beasts to do properly. If you are looking into a kit, look into the linus array.

    Contact Rick craig

    home.earthink.net/~selahaudio
     
  3. Greg Monfort

    Greg Monfort Supporting Actor

    Joined:
    May 30, 2000
    Messages:
    884
    Likes Received:
    0
    Trophy Points:
    0
  4. Kerry Hackney

    Kerry Hackney Stunt Coordinator

    Joined:
    Oct 18, 2000
    Messages:
    206
    Likes Received:
    0
    Trophy Points:
    0
    I was also really interested in line arrays at one point. However, the more I read the more I became convinced that they are extrememly hard to get right. There are a few kits/designs around but of them, most were pretty expensive. FWIW, There is also some info on the Zalytron site about line kits. Also, Bottlehead has one called the Straight-8 I think..
     
  5. Hank Frankenberg

    Hank Frankenberg Cinematographer

    Joined:
    Oct 13, 1998
    Messages:
    2,573
    Likes Received:
    0
    Trophy Points:
    0
    Danny Richie, GR Research, has designed a new line array and sells the kit of drivers (8 BG Neo 8 ribbon tweeters and 9 custom made 7" woofs per tower) and crossovers for I believe $1795. I think it's the first line source kit.
    Here's a link to one guy's description of them:
    http://www.harmonicdiscord.com/forum...=2187&forum=34
     
  6. The linus Array I am talking about cost 500-700 complete kit (don't reember off hand). there was a complete WhitePaper on the project posted on the PEform a while ago.
     
  7. rich r

    rich r Stunt Coordinator

    Joined:
    Jul 29, 2001
    Messages:
    56
    Likes Received:
    0
    Trophy Points:
    0
    Maybe a line array is beyond my design capabilities, but a kit around $500-700 is definitely appealing.

    Anthony any luck on finding the info for that kit you mentioned?????
     
  8. Here are two links for you
    http://www.pesupport.com/cgi-bin/config.pl?read=73504
    which has a bunch of info
    inside that link has the following link
    http://www.prism.gatech.edu/~gte929u/LinusWP.pdf
    It is the Whitepaper on the linus project.
    Jim Griffin has some images (I believe it is him) of a built Linus array. [email protected]
    Also, Rick Craig is offering this array as a kit with matched drivers. http://home.earthlink.net/~selahaudio
    Hope this helps
     
  9. rich r

    rich r Stunt Coordinator

    Joined:
    Jul 29, 2001
    Messages:
    56
    Likes Received:
    0
    Trophy Points:
    0
    anthony,

    thanks for the links they are much appreiated
     
  10. rich r

    rich r Stunt Coordinator

    Joined:
    Jul 29, 2001
    Messages:
    56
    Likes Received:
    0
    Trophy Points:
    0
    Anthony I think the $$'s are the problem
     
  11. John E Janowitz

    John E Janowitz Second Unit

    Joined:
    Oct 30, 2000
    Messages:
    445
    Likes Received:
    0
    Trophy Points:
    0
    Line arrays are something I have been working on for awhile now. Here is a post I did quite awhile ago talking about some designs that were available, and the theory behind the line arrays. I have since done more experimenting with things such as Bessel arrays which I am now very fond of. I'll post more on that when I have time tomorrow, but for now, here is the post I made some time ago:
    Well over on a different forum there was some talk about line arrays. I figure some here might be interested too. I guess I've had enough experience with line arrays to give some meaningful info on the benefits of line arrys. There are several benefits to a line array system over a conventional point source type speaker if it is made right. The main differences are between a conventional speaker, which is a point source, and a line array, which is a line source.
    First off, a line source will typically project it's near field response for a distance of about 8-10 times its length. So an 8ft line array would have a nearfield of 60-80ft. What this means is that within the nearfield the line source will only decrease in output by 3dB per each doubling of distance. On the otherhand, with a typical speaker(point source) will have a nearfield response of only about 1 meter.
    We can compare a 90dB efficient line array to a conventional speaker 90dB efficient. At 1m their responses will be equal. Doubling the distance to 2m will give 6dB less for the conventional speaker, but only 3dB less for the line array. At 4m away, the conventional speaker will be -12dB, and the line array only at -6dB. It is for this reason that line arrays tend to "fill" a room much better.
    The next advantage of a line source is the controlled room interaction. I've often heard that 75% of the problems with speakers are due to room interaction. Reflections off of walls and the floor and ceiling can greatly change the response of a speaker. Where line arrays help is in eliminating problems with vertical reflections from the floor and ceiling. However McIntosh found this only works if you approximate a true infinite line source. This means designing a line array that runs from floor to ceiling to get a true cylindrical wavelaunch. So a shorty 4ft line array would not give these benefits.
    If the line array is also made as a dipole, the horizontal reflections are also greatly eliminated. In a dipole you will have nulls at each side of the line. In a perfect world the front and back waves would completely cancel. In reality this is not possible, but in a dipole line array measuring 40dB less output in the "null" spots is common. What this means is that the reflections off the side walls are virtually eliminated as there is very little output at the sides.
    Another advantage I find with line arrays it that they have no sweet spot vertically. With typical speakers you need to be at ear level with the tweeters to have the best sound. Stand up and they will sound different. With a line array, as long as you are not right near the ends of the line, the output is equal. So whether you are sitting or standing the sound is the same. I find this very important because I like to have good sound whether I'm sitting or standing.
    There are also disadvantages. Unless you can get your drivers very close together your high end response will suffer slightly. This depends on the distance between the drivers. For example we'll take a 4" driver, assuming the edges of the drivers are touching so centers are right at 4" apart. Taking this into account we can find the frequency point where the line will begin to lose directivity, and I believe at approximately four times this frequency(1/4 wavelength frequency) the line will begin to exhibit audible comb filtering and the high end response will begin to decrease. Now, to do this we need to find the frequency with a 4" wavelength. We use this equation:
    speed of sound = freq x wavelength
    Now the speed of sound at STP = 331.45 m/s at STP with dry air, but to get a more realistic number, calculating at 70degrees F with 50% relative humidity the speed of sound would be around 344 m/s. This gives a difference of about 120Hz in this case compared to using the STP value.
    So, the wavelenght is 4" and converting to meters we get .1016 m. Plugging into the equation we get:
    344m/s = freq x .1016 m
    freq= 3391Hz
    This is the point where the line will begin to lose directivity with a 4" driver, and at double this point, 6738Hz, comb filtering will begin to take effect. I haven't been able to do any expermineting yet to determine at what point the comb filtering becomes audible though in relation to this frequency. I have been taking to someone with more real world experience with line arrays, and his would guess say that likely around 13,500Hz, double the frequency where the comb lines begin to take effect they would be come audible.
    The other problem is often the high cost of multiple drivers, and the time required to assemble a line array. Believe me, cutting a total of 48 holes for a bunch of small drivers is not a fun experience.
    All in all if you can find inexpensive drivers to use, I believe the advantages far outweigh the problems. My shipment of 2" x 3.5" line array drivers should be here within the next 2 weeks. I'll give you a little background on these drivers.
    Awhile ago I set out to find an inexpensive driver for line array use. Steve Sedmak had done quite a bit of looking into tall narrow drivers for his line arrays. Typically a tall narrow driver will have very good off axis response. This is a very good thing. So I went out to find some narrow drivers. They needed to have a very smooth frequency response. Inexpensive was a main concern, but they also needed to sound good. My first thought was to find a driver slightly shorter than 4" tall. A 2" x 3.5" tall driver pushed the 1/2 wavelenght frequency up to almost 8000Hz and the quarter wavelength up to almost 16000Hz. Although not that different from the 4" drivers, to me this was acceptable. EQ should be able to compensate for the high frequency rolloff, and above 16000Hz comb lines should be less apparent. When I measured these drivers, I found that they also had a very smooth off axis response, which like I said, was very important. In addition these drivers have a nice response up to nearly 20KHz. They actually measured flatter up to this range than many tweeters I had played with. Finally the nice thing about these 2" x 3.5" drivers is that they are affordable.
    Someone mentioned the Zalytron Axom array on the other forum. As far as I can see there are a few problems associated with the short line of only 4 tweeters. First is the distance between centers on the tweeters. Being apart as far as they are you will still have the problems with high frequency rolloff. Second is the vertical dispersion problem. You really need to be within that range of the 4 tweeters for things to sound right. As you go above or below the 4 tweeters things will sound different. That basically takes away one of the main advantages of line arrays IMO. Having the short line of tweeters will also not give the benefits of canceling reflections from the floor and ceiling as a full line would. Finally is the problem with the rolloff as you exit the nearfield. Like I said before a line typically has a nearfield equal to about 8-10 times it's height. So 4 tweeters would have a nearfield of around 100inches at most. Beyond that the high end will roll off at 6dB per doubling of distance and the midrange drivers would still only roll off at 3dB per doubling in distance. In a small listening room this would not be a problem, but in a large hometheater it could be. This is also a problem seen with their other line array using the Raven tweeter.
    If anyone is interested in any more info on the small drivers, or has any questions on line arrays in general, feel free to contact me.
    Also check out this page http://www.stryke.com/current.htm if you are interested in seeing pics of a few line array's I've played with. None of them use my drivers, but the pics may still be interesting.
    That was the end of my original post. I'll have some more info on the Bessel Arrays, tapering the power at the ends of the line, using ribbon drivers in a line array, and things like that to give tomorrow. I need to get some sleep for tonight though.
    John
     
  12. rich r

    rich r Stunt Coordinator

    Joined:
    Jul 29, 2001
    Messages:
    56
    Likes Received:
    0
    Trophy Points:
    0
    John,

    your post is very informative and resparks my flame of building a line array. I will be contacting you soon

    thanks

    Rich
     
  13. Fred Seger

    Fred Seger Stunt Coordinator

    Joined:
    Aug 13, 2000
    Messages:
    136
    Likes Received:
    0
    Trophy Points:
    0
    John, I'm very interested in seeing what you have to say about bessel arrays
     
  14. James W. Johnson

    James W. Johnson Screenwriter

    Joined:
    May 26, 2001
    Messages:
    1,055
    Likes Received:
    0
    Trophy Points:
    0


    Yeah he quoted me $825 for the kit shipped, it includes drivers and crossover parts only.

    Probably would cost around $1200-$1300 to finish them and I would imagine they would be fairly labor intensive because of the 32 hole cuts that are made on the baffles.

    I have always wanted to try one out and I am sure I will someday but this is more than I would spend.

    Right now my dynamic stereo speaker wish is to find a nice pair of Altec A7s locally.
     
  15. John E Janowitz

    John E Janowitz Second Unit

    Joined:
    Oct 30, 2000
    Messages:
    445
    Likes Received:
    0
    Trophy Points:
    0
    Well, I wanted to hold off a little on commenting about the arrays more because I want some measured data to go with what I have heard, but I can give a general rundown.

    First I'll give you an idea of how the lines were put together. I built a baffle for a line of 24 of my 2" x 3.5" drivers, configured vertically for a total line length of 84". The baffle is open backed to operate as a dipole. The baffle is straight on one edge, and angled on the other. The baffle is about 9" wide at the top and 20" wide at the bottom. I need to get a drawing of this up soon. The drivers are placed 3" from the straight side. Placing the drivers close to one side of the baffle and farther away from the other side is supposed to help out with smoothing out the baffle step. Also the angle on the other side is supposed to help out with smoothing the effects of edge diffraction. I hope to be able to measure these things in the future when I have time.

    Now on to my experiments. First, I wired the drivers up with equal power the whole way down the line. I wired up 4 sets of 6 drivers in parallel. This gave 1.33ohm per group. I then wired these 4 sets in series for an overall impedance of 5.33ohm. When listening to this setup, the comb filtering effects and phasing problems were very apparent in the high frequencies. If you are very close to the line, the high frequencies are apparent, but as you move away, even a 10dB boost in the high frequency range couldn't get the highs to come out properly. The midrange did sound very clear though, and bass was surprisingly strong for a bunch of tiny drivers. Efficiency was very high as all 24 drivers were functioning together. Sound was the same throughout the entire vertical range, sitting, or standing.

    The next experiment was to taper the power, concentrating it more at the center of the line. Jim Griffin talks about this in his whitepaper. I had 24 drivers, so I made 4 banks of series wired drivers. Numbering 1-24 from top to bottom, here is how the wiring went. Drivers 12 and 13 were in series for 16ohm. Drivers 9-11 and 14-16 were in series for 48ohm. Drivers 5-8 and 17-20 were in series for 64ohm. Drivers 1-4 and 21-24 were in series for 64 ohm. Then these were all wired in parallel. If my math is correct, this gets back to an 8ohm load overall.

    Things sounded a little better on this line. Comb filtering and phasing effects were not quite as apparent, but still audible. At the middle of the line, the highs sounded quite a bit better, but were still rolled off. This is because there are now 2 drivers in the middle getting more power than the other drivers. The off axis response of the other drivers is not able to cancel the high frequencies quite as much. If you plan to be at ear level with the center of the line at all times, this may be an option to look into. However, when moving below or above center, the highs instantly disappear again. Efficiency also suffers now because you have the majority of drivers getting much less power. After this experiment, it is my opinion that tapering the power is not the best option for a line array. It lowers the magnitude of the comb filtering effects, but really does nothing to eliminate the problem.

    The next experiment I did was after I already had these groups of drivers in series. I completely removed drivers 1-4 and 21-24 for this experiment. I split up the remainder of drivers so now 5-8 and 17-20 were two different sets, and 9-11 and 14-16 were also 2 different sets. I now wired these driver sets up as a 5 driver bessel array. In the bessel array you typically have 5 drivers, and in this case 5 groups. Drivers 2 and 3 are wired in parallel, in series. Driver 4 is also wired in parallel, but with the polarity reversed. Drivers 1 and 5 are wired in series with each other, and then wired in parallel with 2 and 3.

    What this does is give you one driver in the middle with the driver above, and the driver below out of phase. Then the drivers at the top and bottom are in phase, but getting 1/2 the power. I need to look into the Phillips patent on this to find out how it actually works, but a few things come to mind right away. Driver 2 and 3 are in phase with each other. Typically the off axis response of driver 2 would create comb effects on the center driver as they are more out of phase higher in frequency. However, you take driver 4 and reverse the phase. This will have a vertical off axis response that creates an effect on driver 3 that is equal in magnitude, but opposite in phase to that of driver 2. Therefore the off axis responses of the 2 drivers cancel each other out, leaving almost no net effect on the response of driver 3. You end up not having your high frequencies rolled off.

    At the center of the line, things sounded very good. Moving up and down did sound a little wierd. The sound always appeared to come from the center of the line, which it is supposed to do. If you went up, things sounded fine. Moving down, things sounded a little strange. I believe this is because of the way the center group had only 2 drivers, and the group below that was out of phase had 3 drivers. This line sounded the best just above the center 2 drivers.

    The final experiment I did was to just simply wire 5 drivers up in a regular bessel array. This sounded by far the best of any of the experiments. There were no audible phasing or comb effects in the high frequencies. The sound appeared to come from the center of the line when moving up and down, the response seemed to stay nearly the same. High frequency response did not suffer when going below or above the center as it did in the first tapered power line. There was no audible change when going past the single out of phase driver. Efficiency is better than with the tapered power line. I really like the sound of the line like this. I'm actually going to be making a couple of these 5 driver bessel arrays as computer speakers. Too bad the bessel array is patented or I'd love to do something commercial like it.

    I do plan to do a little more experimenting yet. There is also a 6 driver bessell line. I plan to experiment with that as well. I believe I may be able to get the tapered power method to work a little better using groups in the 6 driver bessell array. I'll have to play with that. Also, I do plan to do frequency response measurements of all of my experiments to be able to see on paper what I have been hearing, and to document it.

    Also, something to mention is with ribbons like the RTW2, the vertical off axis response is extremely controlled to begin with. Drivers of this type will not be plagued by the comb filter effects nearly as much. When I start experimenting with the ribbons for my line arrays I plan to modify the faceplates and make some fins that protrude out between the drivers. This should further help to control to vertical off axis.

    For anyone wanting to experiment in line arrays, I would suggest getting 5 or 6 cheap drivers. Experiment with the different wiring arrangements. The SA2-351 I used for the experiments are $2 each, so it is an inexpensive experiment, but one you can learn a lot from. Hope to have more to share later this week.

    John
     

Share This Page