Why?
I set out to build a line array in 2003. The problem I was solving was that in my studio I needed a second pair of speakers for live rehearsals in which instruments were loud, and PA was needed. Also, for mastering its nice to have at least one other pair of speakers for comparison to monitors. The challenge was to make some very loud hi fi speakers that could handle the stress of PA. Dispersion control was important because while my room is 17x42, it will saturate and make bad low-mid standing waves despite lots of low traps. (3, totaling 64 sq feet of panels … and they make a huge difference) but at a certain gain, the room loses it.
Currently I use a pair of Renkus Heinz SR82’s for PA/tracking monitors. These sound great. One engineer I talked to at Renkus uses these as his home HiFi. They are flat, clean, and have reasonable dispersion control in the horn, but the woofers are more typical of normal cones. They also need a 600 watt amp to breath. They are teamed up with a 3 cube titanic 12” sub through an Audio control crossover @ 70hz. So the SR82’s have two issues: dispersion control below 2400 hz, and needing a big amp. So I needed efficient, loud, hi fi, and controlled dispersion. These constraints point to horn loaded systems or line arrays.
Horns were tempting because they have all that – but I had a choice between making a horn (very hard), buying a hi fi horn good down into the 100-200 hz range (very expensive), and using a pro sound box made for large format – a horn loaded 10 or something which was so much more than I needed, takes up a lot of space, and is expensive. So I was left with Line Array. I will miss the phase alignment of a full range horn system – but that will be… my next project.
Getting Started
Looking at line arrays I studied the work of several diy and semi-pro makers (Jim Griffin etc). Then I looked at
the theory and found conflicts.
Line array design makes much of the near field to far field transition point. While there are formulas for this in several papers, the distance from the source to that point is frequency dependent. In almost all home situations the listening point will see frequencies that are inside AND outside the transition point. Even a floor to ceiling array will have frequencies transition inside ideal range. Floor to ceiling arrays have a nifty quality of reflecting such that the apparent length of the line extends. Cool idea – but it comes with the consequence of floor and ceiling reflections which might be undesirable. In addition LA’s have the obvious issue of comb filtering. There’s also a built in issue with phase alignment because the same sound arrives at your ears from in-phase sources which are different distances from your ears. This issue made me suspect that maybe all the wildly positive testimonials were just people who were smitten with their new toy. The consistent descriptions are “in your face” and “sound stage”. Both of these are describing the impacts one would expect with line arrays. “In your face” because of the near field, and “sound stage” because of the great separation you would get with controlled dispersion and low reflection. To this point I had never heard a small line array so I have relied on the testimony of sources that seem to be reliable.
But…. The good news kept pouring
in. Virtually every respectable pro audio company now makes a line array
system. (Turbosound the exception?
According to its founder the industry is behaving like lemmings, looking for
fools gold etc. He makes his case based on first arrival/phase alignment, and its not a bad argument but somehow I didn’t buy it when
Point being…. If Line Arrays were only pure marketing hype and corporate self interest, the world’s best designers would not be doing it. The designers I know are thoughtful, stubborn idealists, academically inclined, who have no reason to ruin their reputation by participating in a flawed endeavor. Perhaps LA’s are a bit like Transmission line designs in that the theory has a high voodoo ratio and the math is not fully resolved. .. but if they work, why not? What seems most clear is that getting air to move the way we want it to move is still a sticky wicket – and line arrays by teaming up nodes move a lot of air in the way intended by the source material. No honest scientist will tell you we’ve got fluid dynamics licked. The fact that we use cone drivers (or whatever) to move air around is still a fairly pathetic solution relative to the issues or “reproduction”. No speaker can come close to reproducing something as simple as a hand clap accurately – so I’m not about to get all stuffy about “correct” solutions.
Joe D’Appolito (legendary hi fi speaker designer and theorist) is in this game via a hi-fi design for Zalytron, The Zalytron W arrays or Orca Array might have been about right for me – but the woofers are expensive – and a pair of Raven R3’s (the tweeter) costs as much as my entire system. I called them to ask a few questions and it was a big turn off – the people selling the parts don’t seem to know anything about the design – so I just blew Zalytron off. I was asked by several people in forums why I didn’t just go with a proven design. That’s good advice but a) I could not hear any proven designs so I had no way to validate them, b) Most were too expensive, c) the DIY designs were too elaborate to build with my current workshop d) The design issues with line arrays are not particularly daunting as they are with horn loaded systems (so I felt competent to do the design) and e) The cabinets that can be provided with kits are expensive. I’m not saying any of this is unfair, its just that for a person used to diy prices, all that seemed like a big hit.
I already knew I could not route a 7 foot baffle. I knew I
could try, but the consequences of a screw up can be ugly. I have access to an
NC router which I can use to make smaller baffles – but it would not work for
this one. I could have routed it by hand but that’s a lot of mdf dust and I lack a dust removal system. So I decided I
needed to design one for the room and my budget and find someone to help me
build it. I looked for a custom builder who did not have a line array in their
line-up but wanted free “R&D”. I would design it, take the risk, and if it
worked, they could keep the design. The guys at
Driver and Box
Choices
First, I had to decide on the woofers. The obvious choices for a 5” system were the Audax “W” series or the Vifa PL14. Jim Griffin picked the PL14 – and D’ Appolito seemed to prefer the Audax. The Audax was a lot more money, and I have built several sets of speakers with Vifa, so I went with the Vifa. Also, the Linus Array II used the PL14 now – and Version II is always better right?
I had built a set of Monitors with the BG ribbons (Neo3pdr), and the highs are very articulate and transparent. Ribbons look good for line arrays because they are line-source-like. Ribbons more or less make ambiguous the design guideline for Line Arrays that the cone centers should be no further apart than the wavelength of the highest frequency they will reproduce. (That rule is a bit odd by the way, because cone drivers do not by any means reproduce all frequencies from a single point – but, its just a guideline). The trick is getting many tweeters close together, top to bottom so they are more like a line source. This problem was handled by making a template that could mount the neo’s in the well – then the entire template would be loaded and mounted as a single unit. By the way, if there were such a thing as a true line source it might look like a ribbon, but I doubt if it would propagate sound like a ribbon. A true line source, I suppose, must reproduce every frequency from every point on the line. Therein is a shade of the voodoo. Perhaps in the future we will mic violins with a mic that’s the shape of a violin, and reproduce it from an array of ribbons such that each spot in the instrument is represented in the array.
The next important choice was the line length. By line array standards my room is small - Still, the longer the line the better. The near field transition point extends to a reasonable length with a 42 inch line. The next choice was the tweeter line length. This was tougher. BG engineers helped us. BG was already in to building Line Arrays. Good reason too – La’s use a lot of tweeters. They recommended a line of equal or greater length than the woofers because of the dispersion characteristic (not much in the vertical) of the neo3. So we did that. My experience with the Neo’s validated their opinion. They have a big problem with vertical dispersion. All I wanted was something where any audience would be on axis sitting or standing – that came to 42 inches based on measuring myself and some shorter friends. 42 inches puts most possible listening positions in the near field for the midrange – so that was good enough.
Next came the box dimensions. Bass Box Pro told me that 7 PL14’s would be good down to about 60 hz (f3) in a vented box with a flat alignment. I was looking for a speaker that could stand on its own or use with a sub - so 60 is a reasonable compromise. It looked like it might sound decent into the mid 50’s stand-alone - Good enough.
In order that the top tweeter be the same height as my ears while standing the overall height had to be about 6’4”. My height. The front baffle had to be like 11” with the two rows of drivers side by side. I selected 9 for the depth to minimize internal standing waves. Internal structure was similar to the Linus array copy referenced above. Vertical strength would not be an issue with braces between each driver (there are). Overall box size with these constraints was too large for the PL14’s to be happy – so the box has a false bottom. One day I might use that area for an integrated sub.
Crossover
One big issue with a box like this will be the comb filtering between the tweeter and the woofer line. (one that goes away with one big tweeter as in the Zalytron designs). Also, the woofer line had to cross at about 1.5k to meet the cone center/frequency criterion. . The woofer turned out to be flat in the used range – but rise in the last (unused) octave. Given the comb filtering issue, we selected a 4th order filter. I would have preferred 2nd order from a dynamic range/detail pov, but the box promised great dynamic range anyway with all the coupling going on. So, the choice of 4th order values a lot less comb filtering over a teeny loss in detail. Also, since the acoustic roll off was not working with us (rather against us, rising above the crossover point) , 24db electrical was needed. For the high pass filter 24 db was needed for another reason… we were already too close to the lower limit of the ribbon. The BG engineers helped us here too. The Neo3 could handle 1500 hz in a line because of coupling and the presumption that with 12 drivers, each one would never have to work very hard. So, we selected 24 db for the high pass roll off rate and a 1500 hz crossover point.
Some designs offer a tapering of the end woofers to reduce comb filtering – but others do not. I figured I could add this later reasonably well if I needed it.
Chad at
The Box
MDF was the choice for wood – for all the usual reasons. Cherry was the finish. We used a pre-veneered cherry mdf. The factory veneers are solid – better than anyone can do at home due to the high pressures used at the wood fabrications plants. Buried in the overall design were baffle step and baffle diffraction issues.
I did the Baffle-step math (thanks DIY Forum, Bill in particular). The beautiful truth about baffle step issues is that if you have a box you will have issues. It’s not a question of if, but where (hz). With little control over the “where”, and none over the “if”, the right strategy is to soften the transition from 2pi to 3 pi – reducing the suddenness of the baffle diffraction. People do with a curve in the front edges or sometimes oblique angles in the baffle, or off-centering the driver, which, in our case, was happening anyway. The larger the radius, the less abrupt the 2pi to 3pi transition, more or less. How does one make a curve in the corner of a pre veneered mdf sheet?
There are two answers. One is, don’t use a pre veneer – instead, use raw mdf and make a curve with as big a radius as the wood will allow. Then veneer over the curve using a vacuum process. Ad hoc vacuum processes are a pain and also expensive unless you are set up for it. The other way is to use a large matching ¼ dowel to fill the corners. The limit of the radius in this case is smaller. (Actually, there’s a third way involving hunks of hardwood and NC machines, but…). We selected the dowel to stay within the budget. A very small compromise overall.
We’re almost done. The final choices are connectors. I picked Neutrik Speakons because with 19 drivers per side this box could take a lot of power. In theory enough volts to arc with a ¼ inch jack. Bananas would have been fine but speakons are an industry standard, accommodate bi-amping in a single terminal, and have solid conductivity/contact area. Also, they lock. We build external crossover boxes so that tweaking is simple, and Bi-amping does not requiring opening up the box.
Finally