I agree that the CBT approach, using a convex curvature, sounds more natural and does an amazing job of filling the room. However, the focused array has some interesting characteristics that are worth exploring if you are not interested in the constant directivity and coverage.
The speakers in the picture use 21 power amps per side, and the effective curvature can be adjusted from a cell phone app from concave to convex, with a delay range of over 50 inches per driver group. The power amps for the tweeter are only 2W each, and the amps for the woofers are about 15W. That's not enough to energize the neighborhood, but there is enough power to irritate the wife--I've proven that
. Of course, the 250W amps on the 12" woofer help quite a bit.
The details are in this article: 10. Case study #3: A line array with DSP – Audiodevelopers Reborn
This was a prototype to evaluate a number of concepts, but I still haven't been able to do much testing. The DSP in each speaker is controlled by Bluetooth, and you need to connect to each speaker in turn to make changes. I've been trying to redo the interface using the MQTT protocol, which would allow publishing changes to both speakers at the same time. However, that effort is going very slowly.
Also, the effective curvature affects the high frequency response (see the line array analysis thread referenced in a prior post), and I don't have any algorithm implemented yet to change the frequency response as a function of curvature. So it's difficult to A/B the convex vs concave curvatures.
Someday I will probably update this design with better amplifiers and better drivers and take advantage of what I learned in building these. I've become a line array advocate, at least for large listening spaces.
The speakers in the picture use 21 power amps per side, and the effective curvature can be adjusted from a cell phone app from concave to convex, with a delay range of over 50 inches per driver group. The power amps for the tweeter are only 2W each, and the amps for the woofers are about 15W. That's not enough to energize the neighborhood, but there is enough power to irritate the wife--I've proven that
. Of course, the 250W amps on the 12" woofer help quite a bit.The details are in this article: 10. Case study #3: A line array with DSP – Audiodevelopers Reborn
This was a prototype to evaluate a number of concepts, but I still haven't been able to do much testing. The DSP in each speaker is controlled by Bluetooth, and you need to connect to each speaker in turn to make changes. I've been trying to redo the interface using the MQTT protocol, which would allow publishing changes to both speakers at the same time. However, that effort is going very slowly.
Also, the effective curvature affects the high frequency response (see the line array analysis thread referenced in a prior post), and I don't have any algorithm implemented yet to change the frequency response as a function of curvature. So it's difficult to A/B the convex vs concave curvatures.
Someday I will probably update this design with better amplifiers and better drivers and take advantage of what I learned in building these. I've become a line array advocate, at least for large listening spaces.
Attachments
Ongoing thanks again, to all,
Jim, your paper and your posts about your various projects are proving very helpful.
So are comments in the other currently running line-array threads.
My current 'testing room' is similarly sized to your living room, other than 3-4 ft narrower.
It is untreated acoustically for the most part and would be difficult to treat.
If a CBT design could make great sound in it...well, that's enticing !!
I know this is a bit of an aside, but I've found omni type radiators sound better in large reverberate spaces than i would expect.
It's like if you can't beat the room's reverberate nature, join it...with less focused, more diffused radiation.
I say this because my previous listening room was a little larger and i did treat it, getting an even, well damped RT-60 throughout the spectrum.
My omni-type speakers were OK before treatment, but sucked afterwards which was a surprise. Vice versa for the point source type of, which of course was as expected...
That experience, and the way the imaging and the sound-field of the CBT are depicted, are leading me to think maybe the CBT acts as a "direct-omni" so to speak
Neil, very cool project! I have seen it a few times in the past and always admired it....it's probably what planted the seed in my head to try this.
Then the seed got watered by the new beam steering RCF Lol.
Also, I recently got a QSC Q-Sys processor that would make all the DSP pretty easy for the needed channels. You just drag and drop whatever components you need into the processor design. And I could use matrix array presets to instantly switch between different curvatures/shadings/roll offs etc.
Wesayso, always helpful !
TNT, maybe to get started learning, it makes sense to use larger drivers like you did ....gonna keep reading through all the various thread builds
Jim, your paper and your posts about your various projects are proving very helpful.
So are comments in the other currently running line-array threads.
My current 'testing room' is similarly sized to your living room, other than 3-4 ft narrower.
It is untreated acoustically for the most part and would be difficult to treat.
If a CBT design could make great sound in it...well, that's enticing !!
I know this is a bit of an aside, but I've found omni type radiators sound better in large reverberate spaces than i would expect.
It's like if you can't beat the room's reverberate nature, join it...with less focused, more diffused radiation.
I say this because my previous listening room was a little larger and i did treat it, getting an even, well damped RT-60 throughout the spectrum.
My omni-type speakers were OK before treatment, but sucked afterwards which was a surprise. Vice versa for the point source type of, which of course was as expected...
That experience, and the way the imaging and the sound-field of the CBT are depicted, are leading me to think maybe the CBT acts as a "direct-omni" so to speak
Neil, very cool project! I have seen it a few times in the past and always admired it....it's probably what planted the seed in my head to try this.
Then the seed got watered by the new beam steering RCF Lol.
Also, I recently got a QSC Q-Sys processor that would make all the DSP pretty easy for the needed channels. You just drag and drop whatever components you need into the processor design. And I could use matrix array presets to instantly switch between different curvatures/shadings/roll offs etc.
Wesayso, always helpful !
TNT, maybe to get started learning, it makes sense to use larger drivers like you did ....gonna keep reading through all the various thread builds
As I said, I'm interested in updating the amplifier/DSP design, because it is a niche product that no is addressing and I think there might be enough interest in line arrays to make this a viable product. So I might look into working with Parts Express or finding some other collaborator to make a product or kit available. It's not on my short list right now, but it's something I'd like to get to some day.
Even if you could buy a 24-channel amplifier board and DSP software and control app, it would still be a formidable effort to build your own line array. There is a lot of woodworking and wiring involved, plus the cost of the drivers adds up quickly. But these things really sound good, and for a large living room they are ideal. And with built-in WiFi they have very high SAF, since they don't require any additional audio equipment.
Even if you could buy a 24-channel amplifier board and DSP software and control app, it would still be a formidable effort to build your own line array. There is a lot of woodworking and wiring involved, plus the cost of the drivers adds up quickly. But these things really sound good, and for a large living room they are ideal. And with built-in WiFi they have very high SAF, since they don't require any additional audio equipment.
You know, CBT isn't dead !!! I've been thinkin...why not both ??
In Jim's CBT24 thread he says he built the front panel out of 1/2" MDF that bent sufficiently.....I'm saying to myself...hells bells...mount the drivers on the same such a panel, that can be screwed on to either a straight or curved mount.... Anybody see any issues?
I figure I can even make the CBT shading better than usual passive, since I'm already willing to go fully active with however many dsp/amp channels needed for straight line beam steering.
Seems i try both, not so hard...???
Q-Sys... it's awesome, and yes hard to find online info. The guts is a core processor..110f for anything short of a giant installation. There are online level 1 Q-Sys training vids that show the capability. Downer is price...but the 110f's do appear on ebay...i got mine from a bankrupt Toys-R-Us
Thx for the tip on Vituix !
why not both? too much work. I'll bet you stick with whatever you make first
I would do a straight, delay curved CBT iff I wanted on or in-wall mounting but out in the room a physically curved array conveys a degree of immunity from boundary nulls, although I was advised to absorb near reflections for best imaging.
Passive shading isn't hard with Vituix to verify it and show you how well it works. It doesn't seem to be especially critical either. BTW, I modelled both the curved line and its mirror image in Vituix, no other way to include the ground reflection.
I would do a straight, delay curved CBT iff I wanted on or in-wall mounting but out in the room a physically curved array conveys a degree of immunity from boundary nulls, although I was advised to absorb near reflections for best imaging.
Passive shading isn't hard with Vituix to verify it and show you how well it works. It doesn't seem to be especially critical either. BTW, I modelled both the curved line and its mirror image in Vituix, no other way to include the ground reflection.
I would like to add that for a physically curved focussed array you also take advantage of the drivers directivity at higher frequencies besides the absence of phase delay.
Please note that we do not sit located at the focussed points bullet eye, but move the chair/sofa a bit towards the speakers where the listening height is approx. 0,6m (24").
Here we do the measurements and adjustments by MiniDSP and Dirac Live. Deviations in delay is still minor, but realease the listener from being "nailed".
Please note that we do not sit located at the focussed points bullet eye, but move the chair/sofa a bit towards the speakers where the listening height is approx. 0,6m (24").
Here we do the measurements and adjustments by MiniDSP and Dirac Live. Deviations in delay is still minor, but realease the listener from being "nailed".
In Pro audio, the speaker height could differ and hence the need for beam steering, the distance to the listener could vary and hence the need for beamwidth control.
In case of home listening the speaker height is fixed, the listening distance is more or less fixed. Beam steering has no advantage here
Concave CBT. CBT with a twist.
A concave focused array is always criticized for having a very small listening area (point) which is the center of the arc. Here is how to avoid it and make it better.
Make it a reverse CBT. Shade it like a CBT and it becomes a reverse CBT. Dont believe me, just see the polar balloon plots by Don Keeele. The CBT is a constant directivity device even at the rear. A concave array when shaded like CBT is also a CBT array just that it is being heard from the rear.
Now, widen the coverage by placing the focal point very very far (make CBT arc lets say 15 degrees or even less, 5 degrees??) and voila! you just avoided the floor/ceiling and still got constant directivity in the veritcal plane. It has a much wider listening window, vertically.
We now have an array whose vertical listening area (-6dB) keeps getting smaller and smaller with distance but note that its gradual. The focal point is beyond the room by a wide margin so the focal point does not matter.
CBT with a twist!
A concave focused array is always criticized for having a very small listening area (point) which is the center of the arc. Here is how to avoid it and make it better.
Make it a reverse CBT. Shade it like a CBT and it becomes a reverse CBT. Dont believe me, just see the polar balloon plots by Don Keeele. The CBT is a constant directivity device even at the rear. A concave array when shaded like CBT is also a CBT array just that it is being heard from the rear.
Now, widen the coverage by placing the focal point very very far (make CBT arc lets say 15 degrees or even less, 5 degrees??) and voila! you just avoided the floor/ceiling and still got constant directivity in the veritcal plane. It has a much wider listening window, vertically.
We now have an array whose vertical listening area (-6dB) keeps getting smaller and smaller with distance but note that its gradual. The focal point is beyond the room by a wide margin so the focal point does not matter.
CBT with a twist!
Last edited:
CBT is not designed to be constant directivity at the rear. The idea is to generate an expanding wave front with a certain shape and a certain pressure distribution, by locating sources with amplitudes of that wavefront at spatial locations along that wave front. Strictly speaking the directivity of a curved, vertical array only follows the CBT ideas at zero horizontal angle.
Last edited:
The polar plots are symmetrical, ie constant directivity is maintained both in front as well as rear. The idea is to use this symmetry. And yes instead of expanding wavefront its a converging wavefront. If convergence is slow there is no problem.
True, the whole CBT (or straight array talk) is in one plane only, the vertical. So whats the point?
CNC cuts the woodworking down to size
I want one (or two)!
Packaging as an 8 channel plate amp would work well for me. I would buy one for each speaker....
Love the wireless aspect. Not sure how to use it at system level. e.g. I don't think one can select ethernet as an audio output device, maybe I'm wrong; havent tried it.
Oddly enough, cutting all of those holes and rounding them over wasn't that hard. A bit time-consuming, but something you can do while drinking beer, and oddly satisfying. I just used a Forstner bit and drill press, with a router to round the edges. CNC would be a lot faster, of course, but you need an 8' bed and lots of prep. The wiring was much worse, and putting pins on the wires and shoving them into the connectors was painfully slow and annoying.
I used a 20-channel amp for each speaker: 8 woofers and 12 tweeters. Obviously, the design could be simpler for full range drivers.
The wireless aspect is out of scope for this particular forum, but for me it is the only way to design speakers anymore. You can't just use a "WiFi driver", but apps like Spotify, Playcast and even Windows Media Player will allow you to cast the audio to your networked devices. Playcast supports streaming from a FLAC-encoded library, which is what I use. The "downside" of networked audio is the delay, which means it won't work properly with realtime audio from a computer or TV. So for those sources the amp would need an additional input for analog or a low-latency Bluetooth connection.
I used a 20-channel amp for each speaker: 8 woofers and 12 tweeters. Obviously, the design could be simpler for full range drivers.
The wireless aspect is out of scope for this particular forum, but for me it is the only way to design speakers anymore. You can't just use a "WiFi driver", but apps like Spotify, Playcast and even Windows Media Player will allow you to cast the audio to your networked devices. Playcast supports streaming from a FLAC-encoded library, which is what I use. The "downside" of networked audio is the delay, which means it won't work properly with realtime audio from a computer or TV. So for those sources the amp would need an additional input for analog or a low-latency Bluetooth connection.
why not both? too much work. I'll bet you stick with whatever you make first
I would do a straight, delay curved CBT iff I wanted on or in-wall mounting but out in the room a physically curved array conveys a degree of immunity from boundary nulls, although I was advised to absorb near reflections for best imaging.
Passive shading isn't hard with Vituix to verify it and show you how well it works. It doesn't seem to be especially critical either. BTW, I modelled both the curved line and its mirror image in Vituix, no other way to include the ground reflection.
I think I've got how to do both figured out.
It doesn't seem to be too much work, because it appears to me 90% of the labor is in the driver baffle assembly and wiring, which would be the same for either given the muti channel setup (with one exception noted below)
Seems the rest of the labor is just two different boxes to mount the baffle.
The one exception...I'd like a taller straight line than CBT.
So I'm thinking to make the straight line as 2 pieces.
Total straight line of 25 drivers....currently thinking venerable TC9FD.
Main array piece is planned to use 20 drivers which is also for the CBT.
Additional array piece of 5 drivers to lengthen the straight line.
Anybody see any issues with this idea?
(setting aside dsp/amp channel issues)
Thx, mark
well total quantity of work depends upon whether you can bend the baffle after drivers are mounted and its wired. You can certainly do that but it will impact your design in several ways mostly around needing a removable baffle.
I was just picking up pen to post about need for beam steering. In pro-audio, full CBT (vs ground plane CBT) hangs above audience and beam steering is used to optimize audience coverage. In groundplane CBT the apparent point source is a fair distance behind the speaker on the floor and the beam is elevated above the floor at an angle about 2/3 of the CBT arc angle. You just need to be sure the angle is large enough that wherever you listen, your ears aren't above the beam.
When you look at Legendre shading you will see that the top group of drivers are about 20 db down. Leave them out and they won't be missed. What I did (on paper) was start with a 27 driver, 36 degree arc, truncate it down to 30 degrees by leaving out the top 3 drivers and then fill in the remaining 30 degrees with 24 drivers. (IIRC) Now I have to worry if the beam angle is too low but my seated LPs will be out 4m and my standing LPs near the far end of the room. The beam angle is based on the untruncated array's arc angle (correct me if I'm wrong) so I think I'm OK.
You will be more bass shy with shaded CBT than unshaded floor to ceiling array but I'm sure you have plenty of subs
I was just picking up pen to post about need for beam steering. In pro-audio, full CBT (vs ground plane CBT) hangs above audience and beam steering is used to optimize audience coverage. In groundplane CBT the apparent point source is a fair distance behind the speaker on the floor and the beam is elevated above the floor at an angle about 2/3 of the CBT arc angle. You just need to be sure the angle is large enough that wherever you listen, your ears aren't above the beam.
When you look at Legendre shading you will see that the top group of drivers are about 20 db down. Leave them out and they won't be missed. What I did (on paper) was start with a 27 driver, 36 degree arc, truncate it down to 30 degrees by leaving out the top 3 drivers and then fill in the remaining 30 degrees with 24 drivers. (IIRC) Now I have to worry if the beam angle is too low but my seated LPs will be out 4m and my standing LPs near the far end of the room. The beam angle is based on the untruncated array's arc angle (correct me if I'm wrong) so I think I'm OK.
You will be more bass shy with shaded CBT than unshaded floor to ceiling array but I'm sure you have plenty of subs
I'm sure I would screw up doing that "by hand", even without the beer. I prefer to put the time in on CAD for a virtual prototype then CNC with good confidence everything will fit, +/- sanding for tight dadoes.
I see you are thinking an amp per driver. Should be able to do an amp per pair of drivers placed symmetrically about center. For CBT drivers are driven in groups of as many as 12 or 16 at the base down to as few as 4 for the top group. With 24 driver boards you can imagine 3 columns of full range drivers and doing some horizontal beam shaping to keep sound off the side walls.
A little FYI on my Modified CBT24 is that its curved height above the floor is 60.7 inches. But layout the baffle flat and it totals 65 inches before you bend its 0.5" thick baffle to the 36 degrees arc for the 24 drivers. The SB Acoustics SB65 drivers is 2.56" width/height. For my array the center to center spacing that I used is 2.625 inches.
If you are talking about using the TC9FD in a straight line for a either straight array or perhaps CBT array is that it is a much bigger driver (and might I add much heavier unit) vs. the SB65. I have a TC9FD in my hand and it measures about 3.3125" width/height. Hence, each TC9FD is 0.75" larger across than the SB65 per driver. Thus for 24 driver straight array with the TC9FDs you are looking at a total length of that is 0.75" x 24 = 18" longer than my 65" long array. Thus the TC9FD array of 24 is about 83" long.
Now there is the weight differences for the SB65 (0.31 lbs each) vs. the TC9FD (1.54 lbs each). Thus driver weight for 24 SB65s totals to be 7.44 lbs vs. 36.96 lbs. for 24 TC9FD. Those smaller neo magnet driver in the SB65 matter vs. a ferrite magnet unit.
If you are talking about using the TC9FD in a straight line for a either straight array or perhaps CBT array is that it is a much bigger driver (and might I add much heavier unit) vs. the SB65. I have a TC9FD in my hand and it measures about 3.3125" width/height. Hence, each TC9FD is 0.75" larger across than the SB65 per driver. Thus for 24 driver straight array with the TC9FDs you are looking at a total length of that is 0.75" x 24 = 18" longer than my 65" long array. Thus the TC9FD array of 24 is about 83" long.
Now there is the weight differences for the SB65 (0.31 lbs each) vs. the TC9FD (1.54 lbs each). Thus driver weight for 24 SB65s totals to be 7.44 lbs vs. 36.96 lbs. for 24 TC9FD. Those smaller neo magnet driver in the SB65 matter vs. a ferrite magnet unit.
Jim:
Thanks for chiming in. I have a question I'm dying to ask someone who has done it. When the baffle bends, does it compress on the outside of the curve or bend on the inside or both or neither?
I calculate roughly 1/4" difference in length over the length of the arc for a .5" difference in radius. but 1/4" is only .01" over the span of a single driver so its negligible so far as driver footprint goes but I imagine the baffle ends will need to be trimmed after bending.
Thanks for chiming in. I have a question I'm dying to ask someone who has done it. When the baffle bends, does it compress on the outside of the curve or bend on the inside or both or neither?
I calculate roughly 1/4" difference in length over the length of the arc for a .5" difference in radius. but 1/4" is only .01" over the span of a single driver so its negligible so far as driver footprint goes but I imagine the baffle ends will need to be trimmed after bending.
nc535,
On my Modified CBT24 build the baffle and rear panels are the pieces which are bent to follow the arc. The side panels are flat cut to follow the contour dimensions of the arc angle. Hence, the baffle and rear pieces fit inside those side panels. Thus the baffle and rear panels can be trimmed a bit to fit together as needed as you suggested. If you are a woodworking whiz, you can dado cut the inside of the side pieces so that the baffle and rear panel align inside the side pieces. Having CNC is a good thing.
mark100,
You need to decide if you want a straight array or a CBT. I doubt that he can have one line that fits a cabinet for either a straight or curved array. If you want a straight array that is EQ'ed to act like a CBT, then you are talking about time delaying 24 drivers and then amplitude adjusting gains for 3 banks of drivers per side. Lots of circuitry or software to do what a curved CBT configuration inherently does. A little more woodworking complexity with the curved CBT but a more simple electrical/software solution.
Bottom line is that I like the SB Acoustics SB65 as the driver of choice vs. the TC9FD for a curved CBT solution. For a straight array the TC9FD will work OK as shown by wesayso but size and weight differences matter.
Jim
On my Modified CBT24 build the baffle and rear panels are the pieces which are bent to follow the arc. The side panels are flat cut to follow the contour dimensions of the arc angle. Hence, the baffle and rear pieces fit inside those side panels. Thus the baffle and rear panels can be trimmed a bit to fit together as needed as you suggested. If you are a woodworking whiz, you can dado cut the inside of the side pieces so that the baffle and rear panel align inside the side pieces. Having CNC is a good thing.
mark100,
You need to decide if you want a straight array or a CBT. I doubt that he can have one line that fits a cabinet for either a straight or curved array. If you want a straight array that is EQ'ed to act like a CBT, then you are talking about time delaying 24 drivers and then amplitude adjusting gains for 3 banks of drivers per side. Lots of circuitry or software to do what a curved CBT configuration inherently does. A little more woodworking complexity with the curved CBT but a more simple electrical/software solution.
Bottom line is that I like the SB Acoustics SB65 as the driver of choice vs. the TC9FD for a curved CBT solution. For a straight array the TC9FD will work OK as shown by wesayso but size and weight differences matter.
Jim
Last edited:
Hi Jim, hi nc535,
Jim, you may be right that i can't make one line work for both both a straight and curved box. But so far I still have hope.
I've reread your CBT24 build thread, along with the Tech talk on yours and other builds for construction ideas. And other line-array builds here on DIY.
Yesterday, i visited my favorite cabinet shop to look at alternatives for the curved baffle and back plate.
I've built many boxes out of Baltic birch which I adore working with, to the point I really dislike using anything else.
I especially dislike mdf, but more important than a dislike, I don't see mdf holding up moving back and forth between designs.
So, I'm currently looking for a front baffle material that I think could hold up, and curves suitably. Probably will make a laminate out of 2-3 thin sheets of some type plywood.
Even if I get happy with a baffle material, it's easy to acknowledge that the different box implementations may have 'devil details' that derail the dual attempt.
But so far I haven't discovered a knockout blow..so it's smoke-on for now...
The one thing I don't view as a problem is the necessary dsp/channel count.
I have 24 channels, that include individual FIR capability.
I'm thinking to have each driver go to external euroblock connectors...two connectors in parallel to each driver...one to hook to amp, and the other to allow jumping drivers together changeably, externally.
I'm currently planning on 20 TC9FD's for the dual use baffle with 3.3125" center to center....so very close to the height of your 25 SB65's.
Oh, I think the TC9FD's weight is 0.7lb...not 1.54lb (.7kg)...can some verify pls? (You said you have a driver in hand...never want to argue with that
)Would the extra driver weight present a COG issue...for the backwards tilt?
Thx for listening and advice !
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.