That was a terrible lesson to have been absorbed……it matters more than anything else where it counts…..an individual source’s ability to stand alone within a complex environment of sources……and If we apply the same thinking to drivers as ‘sources’…….hopefully the logic can translate.
In the above is the reason for the immense success of the two way bookshelf speaker globally throughout the industry….it sounds good to everyone for a reason……all of the embedded phase issues within a recording get turned around in the crossover region…..mixed together Or homogenized into a nice sound that most folks can get along with…the FRENCH FRY of speakers so to speak….but then listen to a really well designed 3 way……great recordings sound amazing…..but most mass produced recordings sound like
In the above is the reason for the immense success of the two way bookshelf speaker globally throughout the industry….it sounds good to everyone for a reason……all of the embedded phase issues within a recording get turned around in the crossover region…..mixed together Or homogenized into a nice sound that most folks can get along with…the FRENCH FRY of speakers so to speak….but then listen to a really well designed 3 way……great recordings sound amazing…..but most mass produced recordings sound like
Me too!…..but i just use the powered MTM monitors for folks who come in and wanna listen to their stuff loud and over my shoulder…..the 2way MTM mashes up everything and it always brings a smile…..however I’ll gotten it may be. They NEVER get to listen to my Audeze phones though…..those things reveal EVERYTHING. Not a solution I would suggest except for folks who surf only the best of their catalog.
That's a battle I recognize somewhat. The direction I choose within my DSP can get me 2 things.
One - Well optimized stereo with reduced room sound can provide a wonderful experience for well recorded music, it really is captivating to listen to, but it also brings out a worse type of sound/experience with a lot of other tracks.
Two - Focusing on optimizing room response can pretty much hide warts etc. It makes everything sound a bit better, but the best tracks as found in option one loose a bit of their "luster" or "shine". It's the compromise I do tend make to make, to make as much tracks I have and/or like as pleasant as possible to listen to. While I keep on searching to get the best possible compromise between those two.
Just yesterday we, as a family had a lot of fun listening to music together. My son wanted his grandma, 86 years old, to listen to his favorite tracks. She (and we) listened to oldies like Little River Band, Crosby Stills and Nash, Supertramp, Black Sabbath, Rainbow, Ayreon, Infected Mushroom, Thin Lizzy, Frank Zappa and many others. She didn't like Radiohead and a few others (more modern tracks) of my Sons favorites. But my son's long lasting Pink Floyd focus (many tracks) was much appreciated. She's a rocker and has been since it was invented. My mom said: why don't I have this at home. I wouldn't need any TV anymore... just listening to music would be enough.
One - Well optimized stereo with reduced room sound can provide a wonderful experience for well recorded music, it really is captivating to listen to, but it also brings out a worse type of sound/experience with a lot of other tracks.
Two - Focusing on optimizing room response can pretty much hide warts etc. It makes everything sound a bit better, but the best tracks as found in option one loose a bit of their "luster" or "shine". It's the compromise I do tend make to make, to make as much tracks I have and/or like as pleasant as possible to listen to. While I keep on searching to get the best possible compromise between those two.
Just yesterday we, as a family had a lot of fun listening to music together. My son wanted his grandma, 86 years old, to listen to his favorite tracks. She (and we) listened to oldies like Little River Band, Crosby Stills and Nash, Supertramp, Black Sabbath, Rainbow, Ayreon, Infected Mushroom, Thin Lizzy, Frank Zappa and many others. She didn't like Radiohead and a few others (more modern tracks) of my Sons favorites. But my son's long lasting Pink Floyd focus (many tracks) was much appreciated. She's a rocker and has been since it was invented. My mom said: why don't I have this at home. I wouldn't need any TV anymore... just listening to music would be enough.
I’m doing some work with 5 element Bessel arrays currently using 3” fullrange drivers……the initial results are very promising as I’ve never encountered 5 drivers in an array behaving as close to a single point source as this. The desire and work came from the need for a thin horizontal center channel solution to replace the nonsensical horizontal MTM center plagued with fundamental dispersion failures……still folks need a small solution…..and the 5 element Bessel solves almost ALL of the issues…..it’s just that 3” drivers can’t do bass very well………not 5 units anyways.
………..but I think I’ve solved it with a very unusual woofer from Dayton that looks and acts much like a BMR, but isn’t. It can achieve an F of around 80hz in a sealed .15 cuft enclosure and if used up firing with the screen as boundary gain and 90 degrees off axis, won’t need much shaping to work and still lock the midbass to the screen.
………..but I think I’ve solved it with a very unusual woofer from Dayton that looks and acts much like a BMR, but isn’t. It can achieve an F of around 80hz in a sealed .15 cuft enclosure and if used up firing with the screen as boundary gain and 90 degrees off axis, won’t need much shaping to work and still lock the midbass to the screen.
Last edited:
In a Bessel, the net SPL is much less than 5 drivers due to the Bessel coefficients.
what might work better is a horizontal line array on a curved baffle; that would create the wavefront of a point source in the horizontal plane. The curvature would determine the coverage angle. Shade that and you've got a horizontal CBT which might work especially well.
You could do the same thing on a flat baffle by delaying the drive to each pair of drivers to create that spherical wavefront. This is what Danley does in his Jericho horns and Keele in some CBTs.
what might work better is a horizontal line array on a curved baffle; that would create the wavefront of a point source in the horizontal plane. The curvature would determine the coverage angle. Shade that and you've got a horizontal CBT which might work especially well.
You could do the same thing on a flat baffle by delaying the drive to each pair of drivers to create that spherical wavefront. This is what Danley does in his Jericho horns and Keele in some CBTs.
with the woofer support, the 5 element bessel crossed over at 400 hz handles enough power given the equation........net power handling increase 3.5x BUT yes, as you described, efficiency only increases around 1db over a single unit. My model says over 95db max spl so for CC purposes, we're good to go with two of the 6" Dayton mini subs up firing.
The sealed alignment for the subs with an f3 of 75hz works out perfectly with the typical Home Theater subwoofer as well.
The sealed alignment for the subs with an f3 of 75hz works out perfectly with the typical Home Theater subwoofer as well.
I haven't reached conclusions but have two areas where i have made progress in my thinking.
Firstly, concerning Mayhems ideas for me, and thanks so much for those, your knowledge and experience is so far in advance of mine that i will never follow the background of your proposals and thoughts, and the questions others posed about them, and my decision making process for this build and implementation needs me to do that, so, for now at least, i am not going to innovate to that extent, even though the descriptions have some appeal! - thanks again and i hope you will continue to contribute here as i make progress.
So, back on corner line arrays, i know all solutions will be a compromise of some sort, and i don't have extensive experience, but I believe that knowledgeable and experienced people like the half dozen or so of you who have kindly all responded cannot all be kidding themselves! - they must achieve fine reproduction or you would have done something else. i accept and like that!
I can imagine building a rounded off multi-sided MDF (perhaps six sides) vertical box, cabinet, that will sit comfortably deep in the corners and appear satisfactory when aimed in various different ways. this will allow change/tuning of that aspect.
However, despite the extensive answers, i remain uncomfortable with by feeling about phase consistency and the path difference to the ear from the transducers. Everyone seems convincing about the need for correct phase, and considering only the direct path, the disposition of a vertical array of 25 drivers would seem to assure a large variation across the direct wave front reaching the listener. However the first reflection horizontally is said to necessitate attention. By the way many multi driver systems seem to have the same potential problem, if it is a problem!
Then i have also seen curved arrays, convex, further increasing the distance of the more distant drive units. - logic to me says they might be better concave?
Would anyone have the time and patience available to help me with this doubt please?
M
Firstly, concerning Mayhems ideas for me, and thanks so much for those, your knowledge and experience is so far in advance of mine that i will never follow the background of your proposals and thoughts, and the questions others posed about them, and my decision making process for this build and implementation needs me to do that, so, for now at least, i am not going to innovate to that extent, even though the descriptions have some appeal! - thanks again and i hope you will continue to contribute here as i make progress.
So, back on corner line arrays, i know all solutions will be a compromise of some sort, and i don't have extensive experience, but I believe that knowledgeable and experienced people like the half dozen or so of you who have kindly all responded cannot all be kidding themselves! - they must achieve fine reproduction or you would have done something else. i accept and like that!
I can imagine building a rounded off multi-sided MDF (perhaps six sides) vertical box, cabinet, that will sit comfortably deep in the corners and appear satisfactory when aimed in various different ways. this will allow change/tuning of that aspect.
However, despite the extensive answers, i remain uncomfortable with by feeling about phase consistency and the path difference to the ear from the transducers. Everyone seems convincing about the need for correct phase, and considering only the direct path, the disposition of a vertical array of 25 drivers would seem to assure a large variation across the direct wave front reaching the listener. However the first reflection horizontally is said to necessitate attention. By the way many multi driver systems seem to have the same potential problem, if it is a problem!
Then i have also seen curved arrays, convex, further increasing the distance of the more distant drive units. - logic to me says they might be better concave?
Would anyone have the time and patience available to help me with this doubt please?
M
At one single spot/distance? Read the thread: https://www.diyaudio.com/community/threads/infinite-line-source-analysis.314917/ for a little background.
Even if the math is dense, it might explain some of the behavior of line arrays. For instance why they drop in output at higher frequencies. No, it's not only the comb filtering.
The concave, focused array is just that - focused on a single spot, like a magnifying glass. If your usage model is a dedicated listening room with a single chair in it, that might work well for you; otherwise not.
One thing to keep in mind about arrays is that you don't hear the individual drivers, you hear their sum. Their direct waves sum to a single source with frequency, phase, and directivity characteristics according to the array shape and composition. The individual drivers of a line array sum to a continuous line source in the limit as the spacing between drivers goes to zero and as the length of the line array increases. A continuous line source is every bit as ideal as a point source. In some regards it is better: it doesn't have floor reflection nulls, its intensity is invariant with elevation, its intensity varies less with distance from the source than a point source, doesn't have the vertical lobing issues around crossover that multiway designs must contend with.
A finite, discrete line array can't quite reach that perfection. It has a frequency response with some inherent ripple that must be equalized out, it has some response variation with height, it has some vertical lobing in the HF, it has audible combing in the nearfield. Wesayso's implementation shows that all of those compromises can be dealt with to result to achieve an exemplary result. Other accomplished DIYers and engineer/acousticians have visited him and confirmed this result. My own and many others' experience confirms how well a similar line array works even without his refinements. In professional audio, line arrays have become somewhat of a standard for providing audio in large venues (other than movie theaters).
One thing to be careful about when listening to others is confirmation and other kinds of unconscious bias, which is a proven issue in audio that has led to the need for double-blind testing. Many who profess a liking for line arrays have themselves DIYed multiple speaker types and have a wide base of experience from which to make this judgement.
I think we all should be impressed with Mayhem's professional experience and music room. One thing that struck me about his room, addition to how well suited it seemed to be for what he does, is that its very different than yours or mine. In addition to its large size (length), his listening there is all in the nearfield, which rules out line arrays and practically dictates a point source. I know from personal experience that there are just as many compromises inherent in coaxial driver implementation as there are in line array implementation. He accepts those compromises, which work for him in his situation. You can accept the compromises inherent in a simple TC9 line array implementation, because they will work for you in your situation. You can take comfort in the fact that your situation seems to be seated listening at relatively constant listening distance. This usage model defuses line array response variation with height and distance. Your prime 3 seats are distant enough so that any combing effects will be inaudible.
For line array white papers and DIY threads, search "Jim Griffin" and David Smith (SpeakerDave). You might also stumble upon my line array simulation thread from several years ago, which took a random walk through many of these issues, searching for a way to make a line (or other) array more suitable for both seated and standing listening, while at the same time enjoying a simple, unshaded array while both seated and standing.
One thing to keep in mind about arrays is that you don't hear the individual drivers, you hear their sum. Their direct waves sum to a single source with frequency, phase, and directivity characteristics according to the array shape and composition. The individual drivers of a line array sum to a continuous line source in the limit as the spacing between drivers goes to zero and as the length of the line array increases. A continuous line source is every bit as ideal as a point source. In some regards it is better: it doesn't have floor reflection nulls, its intensity is invariant with elevation, its intensity varies less with distance from the source than a point source, doesn't have the vertical lobing issues around crossover that multiway designs must contend with.
A finite, discrete line array can't quite reach that perfection. It has a frequency response with some inherent ripple that must be equalized out, it has some response variation with height, it has some vertical lobing in the HF, it has audible combing in the nearfield. Wesayso's implementation shows that all of those compromises can be dealt with to result to achieve an exemplary result. Other accomplished DIYers and engineer/acousticians have visited him and confirmed this result. My own and many others' experience confirms how well a similar line array works even without his refinements. In professional audio, line arrays have become somewhat of a standard for providing audio in large venues (other than movie theaters).
One thing to be careful about when listening to others is confirmation and other kinds of unconscious bias, which is a proven issue in audio that has led to the need for double-blind testing. Many who profess a liking for line arrays have themselves DIYed multiple speaker types and have a wide base of experience from which to make this judgement.
I think we all should be impressed with Mayhem's professional experience and music room. One thing that struck me about his room, addition to how well suited it seemed to be for what he does, is that its very different than yours or mine. In addition to its large size (length), his listening there is all in the nearfield, which rules out line arrays and practically dictates a point source. I know from personal experience that there are just as many compromises inherent in coaxial driver implementation as there are in line array implementation. He accepts those compromises, which work for him in his situation. You can accept the compromises inherent in a simple TC9 line array implementation, because they will work for you in your situation. You can take comfort in the fact that your situation seems to be seated listening at relatively constant listening distance. This usage model defuses line array response variation with height and distance. Your prime 3 seats are distant enough so that any combing effects will be inaudible.
For line array white papers and DIY threads, search "Jim Griffin" and David Smith (SpeakerDave). You might also stumble upon my line array simulation thread from several years ago, which took a random walk through many of these issues, searching for a way to make a line (or other) array more suitable for both seated and standing listening, while at the same time enjoying a simple, unshaded array while both seated and standing.
Its not the center to center distance between drivers that determines where wavefront interference begins but the difference in path length from each driver to your ear or the microphone. If you stand on a ladder, sit on a high horse, or otherwise place your ear above the array, then the path length difference will indeed be equal to the CTC distance and you will observe nulls at the frequency where the CTC is half a wavelength.
However, come down off that virtual high horse and seat yourself several meters out into the room at your primary listening position and an entirely different picture emerges. In that position, there will be one driver at your ear height and one driver 100 mm or so above it. Draw a right triangle where the base is the distance from your ear to that driver at ear level and the hypotenuse is the line from the driver 100 mm up. According to Pythagorus, the path length difference is now sqrt(D*D + CTC*CTC) - D. Let D = 3m and CTC = .1m. PathLenDiff = sqrt(9.01) - 3 =1,6 mm. The frequency at which wavefront inference effects begin at the listening position has been raised by a factor of about 160 - out of the audible range!
A typical line array is roughly 2m tall. If we measure and equalize at its vertical center, then the CTC from center to end driver is roughly 1m and the PLD =sqrt(10) -3 = 160mm. This is well within the audible range but this is only one driver among many. Each driver in the array has a different PLD. When we take that into account, we get some response ripple, which is most easily seen in a Vituix simulation model, as I showed in my white paper. This ripple can be equalized away but the equalization diminishes in effectiveness with movement away from the measurement location.
Thanks so much for your continuing patience and repetitions for me. I feel these explanations underlined in the quote are what i have been missing or misunderstanding. Next question, if i may, when does the line array that sums as you describe fall back to a finite, discrete array? - and why.
i recall reading earlier about transducer spacing but guess this is also about the height of the array in between the floor and ceiling?
M
I’ve been involved with the development and design of commercial arrays for over two decades…..i‘m very familiar with how they perform in real world situations……and not sure where you’re gonna get the ‘several meters’ of listening distance from in most homes…..the distance these things need to sum and how they transition to the far field is well understood.
My objection to all of these recommendations is the full range driver solution…….the by far worst implementation of a line source alignment….by far. Add a row of planar drivers next to the TC9s and we’re in a completely different arena of performance………without question. By design…..choose the right drivers…..easy to implement….cost wise….a different story. The TC9 is no longer the budget gem it was at less that $10 a pop pre Covid now over twice that…..and now there’s 8” long planar drivers from GRS at a fraction of what line source tweeters used to cost……Wanna build a home array?……..do it as a 2 way……shorten the line to 16 woofer elements to stave cost……..the floor/ceiling thing is being waaaaaaaaay over blown here………..
Last edited:
If you simulate those planar device line arrays as I have done, you see they fall short for similar reasons to the cone drivers. With planars, its the gap between adjacent planar radiators that matters instead of the CTC. With planars, you need that gap area to be <=10% of the radiating area. (This comes from an AES paper) With the GRS devices, there is too much framework around the radiating surface for this to be possible. These devices no doubt have a nicer treble than the TC9s but they fall short of the ideal line source behavior in the same way that the TC9 array does. I don't know to what extent this compromises the sound; only that it is compromised. Only your ears can tell you if that is a compromise you can live with. And then you have to figure out how to add the bass it needs. This would not be a nestle_into_corners solution.
Personally, I would (tempted to ) go with an array of 48 Dayton DMA-45 1.5" drivers. I reviewed and updated my simulation of it this morning and it looked pretty good but I didn't get to the point of reviewing what it needed in the way of woofers/subwoofers. IIRC it needs to be next to a line of e.g. 5" woofers. If one had room for that, it would really shine.
Personally, I would (tempted to ) go with an array of 48 Dayton DMA-45 1.5" drivers. I reviewed and updated my simulation of it this morning and it looked pretty good but I didn't get to the point of reviewing what it needed in the way of woofers/subwoofers. IIRC it needs to be next to a line of e.g. 5" woofers. If one had room for that, it would really shine.
when I say "finite", I mean it has limited length, as opposed to a line of infinite length, which is often analyzed and used as a standard for comparison
when I say "discrete", I mean the line array is composed of a number of separate drivers at some spacing, as opposed to a theoretical line source that vibrates as a unit along its entire length.
All practical arrays are finite and most of them are discrete. There is no falling back, finite and discrete are what we have to work with. The question is how well do our finite discrete line sources approximate the theoretical, infinite length continuous line source. there is a range of frequencies over which we have a good approximation - from upper bass to mid treble or so. If you want better treble, use more smaller drivers. If you want better bass use bigger drivers. These two things are at odds so you may end up with a bass line and a treble line. To avoid that, design a single line to work down to 80 hz or so and augment with subwoofers.
The TC9s do remarkably well but most of us who use them have some loss of HF hearing due to age and perhaps too much loud music in the past. Someone with full range hearing might not be satisfied with TC9 treble. I think we can all name one person who isn't
I definitely fall in the "most of us" category, i have recently been plotting frequency response from individual test tones, and although the gap between frequencies is quite large up there, i know where the audible tone goes away for me...........however, there are young people and dogs to consider as well!
So, planning for TC9's, - can the units practically be mounted at 83.7mm centres, i.e. face plates touching, if so for Height, i could use 28 x83.7 =2343.6, if they really butt together, room is 2400, just 56 mm left, half each end, is that a benefit over the 25 or does that take the top end of the line into some reflecting mess behind the cross beam and is not worthwhile? With 25, the ends would terminate short of the floor and ceiling by a small amount.
So, planning for TC9's, - can the units practically be mounted at 83.7mm centres, i.e. face plates touching, if so for Height, i could use 28 x83.7 =2343.6, if they really butt together, room is 2400, just 56 mm left, half each end, is that a benefit over the 25 or does that take the top end of the line into some reflecting mess behind the cross beam and is not worthwhile? With 25, the ends would terminate short of the floor and ceiling by a small amount.
I’m not sure which planar you’re referring to……the PT6816-8 element has a length of 182mm and an overall length of 200mm……..that’s more than enough to develop a cohesive line source in multiples. Used in multiples of 8 or more, in home use, allows for a very low crossover point of 1k……or take advantage of the narrow element and cross as high as you like…..4K isn’t unreasonable with the TC9
I don't have drivers in hand but went by the data sheet drawing.
the data sheet shows on overall height of 200 mm and requires a cutout of 175mm which is already less than your 182mm. The height between topmost and bottom-most openings in the metal grill (I assumed it was metal) is a bit less. I scaled the drawing with vernier calipers and took a proportion to come up with what I did. I don't have that final number in front of me but just considering the 175mm cutout, its very different from yours. One of us has an out of date data sheet.
the data sheet shows on overall height of 200 mm and requires a cutout of 175mm which is already less than your 182mm. The height between topmost and bottom-most openings in the metal grill (I assumed it was metal) is a bit less. I scaled the drawing with vernier calipers and took a proportion to come up with what I did. I don't have that final number in front of me but just considering the 175mm cutout, its very different from yours. One of us has an out of date data sheet.