Building a Line Array, what should I be concerned about?

I am building a pair of Line Array speakers for some mild PA use here and there (mostly because I am tired of the "blaring" sound inherent in almost all modest sized PA speakers when they are pushed at all). I already have all the parts I think I will need. A pair of amps with plenty of power, a Behringer Electronic 3-way crossover (might use a separate subwoofer in some cases), an Alesis dual 32-band EQ, 10 6 1/2" Coustic Design Reference woofers (that apparently were made by Eminence) per side, and about 12 Peerless 1" silk dome tweeters per side. These are all very good drivers, but I am not interested in making these sound flawless with super high end drivers, I am just shooting for sound that doesn't sound like the speakers are trying so hard. I have been using a pair of JBL 15" 2-way PA speakers most of the time, and they sound OK at low to medium volume, but just sound unbearable at levels that I consider to be expected SPL levels from them.

My concern is that I have never actually done a LIne Array of my own before, and I remember hearing that comb filtering and other issues can arise in certain situations, and i was hoping to get some preemptive feedback so I don't get to the end of this project and have to go back to the drawing board right away. I want to make these this week, and plan to document my progress with photos on this thread, so I hope somebody out there with L.A. experience can give me some "rules of thumb" before I get started.

Any thoughts?
 

gtforme00

Member
2004-10-21 4:35 am
Read that entire paper as many times as you need to comprehend it. Consider that your introduction to line-arrays. Once you comprehend that paper, you will know the majority of what you need to know about building a practical array. If you still care to learn more, look up the references from that paper, particularly the Urban and Ureda references. You will want to remember that you are most certainly going to be in the far field for the majority of the frequency spectrum, and that Jim's paper is focused towards near-field applications.

Here is a summary though. I'll break it down to you as bluntly as possible.

With 6.5" midbass drivers and 1" dome tweeters presumably having a faceplate larger than 1" in diameter you are not going to be able to meet the criteria for driver spacing suggested by those papers.

With 6.5" midbass drivers, comb filtering is going to start in around 2100Hz in the far field.

You will want to set your crossover point below 2100Hz.

With 1" tweeters on 3" center to center spacing, comb filtering will start in around 4400 in the far field. You are going to run into trouble here if you need to output frequencies higher than that. If you want to avoid comb filtering at the higher frequencies, you will want to get those tweeters closer together. If I were building the array, I would also use more than 12 tweeters per side, as the crossover point will be rather low and you will need the extra output capability of more tweeters to help keep the distortion low.

So what to do?

Since you already have the drivers, you might be tempted to go ahead and throw them together into a line array. If you do, keep the space between each driver and the space between the woofer and tweeter array at a minimum. Crossover as low as the tweeters will allow with high order crossovers (you did not mention the slope of your electronic crossover). You may end up with something better than your 2-ways, you may not.

Remember that EQ'ing comb filtering is not possible as the nature of comb filtering is to create high and low SPL zones in a comb like pattern. If you boost a certain frequency to make up for a low in a given part of the room, you will create a hot spot in some other place in the room. You can still EQ for room effects.

If you decide to pick other drivers, Try to find tweeters with the smallest flange and the lowest resonance frequency possible. If you could find something with less than 1.5" diameter flange and less than 1.5k resonance, you will be able to push the onset of comb filtering up to 10k or so where it doesn't matter so much.

Good luck with your design, feel free to ask for more help.

Regards,
David
 
I have read through the paper twice, but your comments have probably done more for me.

I do have these drivers in a square flange also, so I presume that would be preferred to keep them close together.

I was hoping to cross everything over at somewhere between 1800 and 2000, so hopefully that will help keep the woofer comb filtering in check, although I am a bit confused about where my tweeter troubles will be.

My crossover is This Behringer
 

gtforme00

Member
2004-10-21 4:35 am
Do you have a link to the spec sheet for the tweeters? If you can tell me the flange dimensions, I'll tell you where the trouble begins. Or you can calculate it yourself using this formula.

Wavelength = 13600/d

13600 (speed of sound at sea level in inches per second)
d = diameter (or length of shortest flange dimension) in inches

At the frequency calculated by this equation, the effects of "comb filtering" will begin to influence the output pattern. See Figure 7 in Dr. Griffin's paper. The comb pattern begins gradually and the peaks and nulls become increasingly distinct as the frequency rises. The pattern maximizes at twice the frequency calculated (the first cancellation frequency). This is why you want to push the frequency up as high as possible, to get that first cancellation well above the operating frequency.

All of that is a bit theoretical however, and I will disclose something to you that I have not seen mentioned in line array discussions. As long as the individual array elements are in phase (in other words they are driven by the same source without individual time delay, which is probably all home audio line arrays), and the drivers are not theoretical point sources (which none of the audio drivers I've ever seen are), the cancellation will not be as bad as theory predicts.

Let me explain. The ideal (for our audio purposes) line array is one long thin rectangular element that is moving in phase. If you take that long element and cut it into 12 different smaller rectangles that have no space between them, then as long as they are all still in phase, there will be no difference in output. (This does not apply to phased arrays, which are a totally different beast.) Now take those 12 different smaller rectangles and round them off into circles but keep them almost touching. You now have a compromised array, because you have introduced gaps due to the circular area of the circular drivers not filling the entire rectangular array. However there still is some active radiating area between the centers of the drivers, so cancellation won't be quite as bad as theory predicts. This is why planar tweeters have different rules in Dr. Griffin's paper, and notice how it is the active area that is deemed important in those applications.

Here is the deal though: the theory still gives you the rules to follow. Keep elements as close together as possible, with as much active radiating area as possible for a given rectangular array. It is difficult to predict the results you will have once you stray from the rules set up by the theories. But that is where the yourself comes from in DIY audio!

As for me, I would love to build an (audio) line array one day, as I have had experience designing SONAR ones. Yes, that is right, I've never built a line array for audio, but I have a good grasp of the theories and practicalities of line arrays. I don't even know if an array is going to work well for your PA application, but I see the advantages of controlled directivity and would wager in that direction. Take my advice for as much as you value it, but listen to the experience of people who have actually built and used arrays.

Oh, and a fourth order crossover is probably the minimum you would need, which happens to be exactly what you have in that Behringer unit, so I would say stick with it unless you have troubles!

Best regards,
David
 
Opinion: for PA work, using lines with reasonably-spaced drivers will produce a comb/lobe/beam that is usually, at the worst, only a very minor source of irritation and only for very few listeners. Like only those people who try to listen for such things....like diy-ers out for a show. Excessive loudness, clumsy mixing, piercing high-end distortion are far greater causes of listener unhappiness.

The vertical placement can make a big difference. I think putting the line up high takes advantage of room effects (ceiling effects in a small hall), and that seems to erase-out some of the problem. If I am at a convention or speech set in a big hall, a situation where I can really hear the sound system at work, when it sounds great no matter where I am...most likely they have line-like arrays flying way up high.