with 3D printing you can design much better the conical to rectangular throat adapter, with constraints for equall path length and tapering. you need quite some 3D design skills. sketchup will not do the job...
I am considering building a 8ch dsp amp (similar do the nanaodsp board), so the delays can be designed to your linking as well as the shading
I am considering building a 8ch dsp amp (similar do the nanaodsp board), so the delays can be designed to your linking as well as the shading
I recommend using the Morel MDM 55. Sensitivity is much better than with the Aurasounds and it is not that much larger.
Let me say first: I never heard my prototypes with a stereo setup. So I can't judge how good they really sound. They were just prototypes.
But the in-wall speaker we finally developed will be installed this year. Then I can tell more about how they sound.
So I built one of these.
I have a pile of failed projects that I've never posted, and I thought this was one of them. I built it about six weeks ago, I measured it, and the measurements weren't so hot.
So I didn't bother to post anything online.
I put the thing on my desk to listen to it, and it's stayed there ever since. It's really remarkable how 'listenable' this thing is.
The thing I noticed in the first couple weeks was that it has really good midrange. I've built many Unity horns, and one of the places that I often stumble is when designing the crossover; if you're not persistent it can be challenging to get the midrange transition right.
This "CBT + waveguide" thing doesn't suffer from that; there's no crossover in the midrange, It Just Works.
After a couple weeks I liked it enough that I added an EQ into the mix to flatten out the response. That improved the sound.
Around the same time, I flipped the array on it's side, to see if it would work horizontally instead of vertically. It actually worked better, so that's good. On it's side, the treble is a little more consistent and extended.
My "clone" of FoLLgoTT's design is using this driver:
Tectonic Elements TEBM35C10-4 Miniature BMR(R) Driver | Medley's Musings
This driver has exceptionally low efficiency. I personally believe that most drivers are limited by xmax, and this driver has a lot. So my thought is that this driver might perform better than expected, because it has so much displacement. This turned out to be true; I am listening to the speaker as I type this, and while listening to it play a podcast, I can actually feel the speaker shake the desk (with voice!) And my desk is built like a tank, it weighs about 200lbs. These little drivers can move some air if you use enough of them.
About the only "X Factor" here is that the treble doesn't sound like a tweeter does. I need to do some more research on this. The high frequencies are equalized to be flat, but they don't sound like a dome at all. It doesn't sound "bright" it actually SOUNDS like it's rolled off a bit, but it's not. This is probably some directivity artifact that I need to research. (Perhaps the power response is falling?)
But I'm not complaining here, this thing is REALLY easy to listen to. In fact, the main reason I haven't built a second one is because I'm now leaning towards building a BIGGER one.
Here's the phase and frequency response, measured ground plane outside. Like I said, not a good looking measurement. The phase looks great. These things have that 'articulation' thing going on that Unity horns and full range speakers do real well. I listen to a LOT of podcasts, I basically have them on 40 hours a week while I write software. Being able to get the midrange right is really important to me, far more important than bass or treble.
Here's the frequency response and distortion. I'm using power tapering on the CBT, so it was surprising how loud this thing could get.
Last but not least, I forgot to mention:
I started out with the power tapering set up sothat the loudest drivers were in the CENTER of the array. I eventually switched to what Keele does, where the driver at the END of the array has the most power. The latter works better; it measures better and it also allows for a wider stage. (My CBT waveguide thing is on it's side, not vertical like FoLLgoTT)
I have a pile of failed projects that I've never posted, and I thought this was one of them. I built it about six weeks ago, I measured it, and the measurements weren't so hot.
So I didn't bother to post anything online.
I put the thing on my desk to listen to it, and it's stayed there ever since. It's really remarkable how 'listenable' this thing is.
The thing I noticed in the first couple weeks was that it has really good midrange. I've built many Unity horns, and one of the places that I often stumble is when designing the crossover; if you're not persistent it can be challenging to get the midrange transition right.
This "CBT + waveguide" thing doesn't suffer from that; there's no crossover in the midrange, It Just Works.
After a couple weeks I liked it enough that I added an EQ into the mix to flatten out the response. That improved the sound.
Around the same time, I flipped the array on it's side, to see if it would work horizontally instead of vertically. It actually worked better, so that's good. On it's side, the treble is a little more consistent and extended.
My "clone" of FoLLgoTT's design is using this driver:
Tectonic Elements TEBM35C10-4 Miniature BMR(R) Driver | Medley's Musings
This driver has exceptionally low efficiency. I personally believe that most drivers are limited by xmax, and this driver has a lot. So my thought is that this driver might perform better than expected, because it has so much displacement. This turned out to be true; I am listening to the speaker as I type this, and while listening to it play a podcast, I can actually feel the speaker shake the desk (with voice!) And my desk is built like a tank, it weighs about 200lbs. These little drivers can move some air if you use enough of them.
About the only "X Factor" here is that the treble doesn't sound like a tweeter does. I need to do some more research on this. The high frequencies are equalized to be flat, but they don't sound like a dome at all. It doesn't sound "bright" it actually SOUNDS like it's rolled off a bit, but it's not. This is probably some directivity artifact that I need to research. (Perhaps the power response is falling?)
But I'm not complaining here, this thing is REALLY easy to listen to. In fact, the main reason I haven't built a second one is because I'm now leaning towards building a BIGGER one.
Here's the phase and frequency response, measured ground plane outside. Like I said, not a good looking measurement. The phase looks great. These things have that 'articulation' thing going on that Unity horns and full range speakers do real well. I listen to a LOT of podcasts, I basically have them on 40 hours a week while I write software. Being able to get the midrange right is really important to me, far more important than bass or treble.
Here's the frequency response and distortion. I'm using power tapering on the CBT, so it was surprising how loud this thing could get.
Last but not least, I forgot to mention:
I started out with the power tapering set up sothat the loudest drivers were in the CENTER of the array. I eventually switched to what Keele does, where the driver at the END of the array has the most power. The latter works better; it measures better and it also allows for a wider stage. (My CBT waveguide thing is on it's side, not vertical like FoLLgoTT)
Last edited:
Hi Patrick,
Eager to know a few details. Pls share a few photos of vertical as well as horizontal array. How did you implement the active part and delay? Whats the vertical size, horizontal size, CBT beamwidth angle and waveguide beamwidth angle? For vertical array, it looks like you are using a ground plane CBT, the desk serves as a reflection which doubles the effective height and lowers the pattern loss frequency, both good things. For horizontal array, its not clear how its a CBT, since you said you are using the end driver with the highest power (ie a ground plane CBT) but with no ground plane (or side wall) to give the reflection.
Thanks
Eager to know a few details. Pls share a few photos of vertical as well as horizontal array. How did you implement the active part and delay? Whats the vertical size, horizontal size, CBT beamwidth angle and waveguide beamwidth angle? For vertical array, it looks like you are using a ground plane CBT, the desk serves as a reflection which doubles the effective height and lowers the pattern loss frequency, both good things. For horizontal array, its not clear how its a CBT, since you said you are using the end driver with the highest power (ie a ground plane CBT) but with no ground plane (or side wall) to give the reflection.
Thanks
I have never, ever listened to a live performance and thought: hey, this top end sounds like a dome tweeter.
In fact, after playing with arrays for quite a while I don't think I want my top end have a sound like a dome tweeter. I've listened to a small near field setup with Pro monitors and was amazed at how much high frequency fuzz there was. It wasn't distracting as such, I just needed some time to get used to it again. But it never struck me as sounding right.
The coherence an array is capable of is way more important to me, personally. However FIR correction might bring your top end into play. If the driver has a clean top end, as you get what you put into an array back on steroids.
Last edited:
I need to do that. It looks very similar to OPs
I physically curved it. I designed it in 3D, printed it in 3D, and each element is 'splayed' by six degrees. A total curvature of 30 degrees.
The laptop I designed it with was stolen by a hobo so I'll have to do these numbers from memory:
Vertical and horizontal size is about 16" x 16", similar to the waveguide in my old Summas. Vertical beamwidth is 30, horizontal is 90, but it's currently flipped on it's side. The CBT beamwidth is 30. (five units, each unit splayed six degrees.)
It's in my office and my office desk is in a nook. So the waveguide-cbt thing is stuck in a corner.
I have never, ever listened to a live performance and thought: hey, this top end sounds like a dome tweeter.
In fact, after playing with arrays for quite a while I don't think I want my top end have a sound like a dome tweeter. I've listened to a small near field setup with Pro monitors and was amazed at how much high frequency fuzz there was. It wasn't distracting as such, I just needed some time to get used to it again. But it never struck me as sounding right.
The coherence an array is capable of is way more important to me, personally. However FIR correction might bring your top end into play. If the driver has a clean top end, as you get what you put into an array back on steroids.
Yeah it's really disconcerting. TBH, my Summas did that too, and I'd always assumed that it was because they roll off at 16khz. I figure that the missing output from 16-20khz was that "sparkle."
But this array goes out to 20khz, has no sparkle.
I wonder if constant directivity loudspeakers might require a 'tipped up' treble?
Interesting stuff. Carver is now using power tapering.
YouTube
discussion begins at the 0:10 second mark
Carver mentions (somewhere in those 5 videos i think?) that the drivers are his design and that he got a company in china to manufacture them on agreement that he has the exclusive rights to them for 2 years, but after this they can sell them under their own name.....
The video is a year and a half old
The bass/mid driver looks very similar to the one you use in your CBT.
Can you put the measurements for your CBT in perspective against other designs? For instance, distortion: lower than a good compression driver? power response?
Do you mean the tweeters? They look like Dayton PT2C-8 (description: 'Perfect for use in line arrays') which already are on the market for quite some time.
It may be interesting, but i found the image confusing. I spent lots of time listening to his arrays last capital audio fest, changing my seating position...but the image was always somehow phasey. Something iritating was happenig and my brain did not like it.
May be placement of those midranges. Forget the tweeters for a moment and visualize how the mids project sound. Towards each other, and towards the walls. Not ideal.
I think in an effort to make arrays very narrow, he made bad decision on mids.
Just my personal view.
May be placement of those midranges. Forget the tweeters for a moment and visualize how the mids project sound. Towards each other, and towards the walls. Not ideal.
I think in an effort to make arrays very narrow, he made bad decision on mids.
Just my personal view.
Carver mentions (somewhere in those 5 videos i think?) that the drivers are his design and that he got a company in china to manufacture them on agreement that he has the exclusive rights to them for 2 years, but after this they can sell them under their own name.....
The video is a year and a half old
The bass/mid driver looks very similar to the one you use in your CBT.
Can you put the measurements for your CBT in perspective against other designs? For instance, distortion: lower than a good compression driver? power response?
I've never been a fan of arrays. What Adason describes has been my same reaction. All the line arrays I've heard until 2014, they always made everything sound huge and the image was unfocused.
Y'know how a good set of two way monitors can throw a pinpoint stage? Line arrays are basically the opposite of that.
But the CBT is different; it's not as pinpoint as a minimonitor, but not as diffuse as a line array.
It basically falls in the middle of the two presentations.
The surprising thing to me, is how loud my little CBT+waveguide thing gets. It only has five drivers, and I figured it would distort. But it's really well behaved; it gets as loud as I'll ever need for speakers in my office. I haven't tried it in my living room yet.
It's really difficult to choose between the two options. My unity horn projects have more "sparkle" in the upper treble. It's really really difficult to generate much output above 10khz with a CBT. If you look at the CBT36 user's guide, it's efficiency at 10khz is something like 80dB, and that's with 144 tweeters!
But yowza is the mid-range great on these things. Unities have really nice mid-range too, but I think this CBT + waveguide thing may be a tiny bit better.
I wonder if any of this has to do with HOMs? When gedlee summas were what I listened to, I could listen for hours on end. These speakers are like that too.
Last but not least, CBTs are hard to screw up. My entire crossover is two resistors.
With the treble output, I have really changed my tune in the last couple of years. I just dont think that 14kHz + matters much to the sound/dopamine calculation at all. My current jbl2452H titanium diaphrams break up at around 15kHz, and although I can hear it, it really makes shockingly little impact on the sound quality, but the graph looks truely awful!
Being in my 30's I can still hear out to 17kHz, but I have even tried to eq the breakup up by 10db, (just to see) and it really wasn't *that* bad. It just sounded like the old focal/JM Labs toxoid tweeter... Some sibilance, but otherwise most would call the sound *exciting*.
When in doubt, its good to remember that records don't play that high, and I don't hear many complaining about that.
Coming from someone who has made as many projects as yourself, thats quite a statement!
I think the biggest advantage is that the CBT speaker could be considered high on the WAF scale (when done right) especially compaired to the other options I would consider to be modern attempts at Hi Fidelity.
I anecdotally imagined that CBTs may have the kind of advantage that the AMT tweeter has of not having to work hard.
With pattern control I think this is the next frontier. Not gaining it, but deciding what the ideal amount of direct vs reflected sound is. Or, making speakers that give you control over the proportions. I notice that Linkwitz keeps his eye on the competition fairly sharp with this page:
Constant directivity loudspeaker designs
He has high praise for the Beolab 90!
On the HOM front- I can't remember where we got with this one. Audiable or not? Controversial if I remember right.
The Achilles Heel of arrays is that the maximum output at high frequency is that maximum output of a single unit. So you can use 144 tweeters like the CBT36 does, and the maximum output above 10khz is still something like 85-90dB. It's really low.
I still need to tinker with them so more, but I think that's why my CBT-ish thing lacks "sparkle", it just can't generate much output above 5-10khz.
An externally hosted image should be here but it was not working when we last tested it.
My local Best Buy has the big expensive CBTs set up for it's sound system, and the lack of output above 5-10khz is unmistakable.
I have some ideas on how to fix this; I think that a long line of small tweeters may not be the ideal solution. (stay tuned.)
They really are special. I can see why Keele has practically abandoned all other technologies.
One of the really interesting possibilities I think, is to use one technology on one axis and another technology on the other. For instance, this thread is about a speaker that's a CBT on the Y axis and a waveguide on the X axis.
Ulrich Horbach, who worked with Keele on the Horbach Keele filters, has designed a speaker that's a WMTMW on the X axis and a shaded array on the Y axis.
I honestly think these type of combos are likely the future. If you do a CBT on both the X and the Y axis, the size just gets insane fast.
Beolab 90 is the best speaker I've ever heard. They (finally) opened up a B&O store in Orange County, I need to get out there and hear the Beolab 50. I tried to hear the Beolab 5 in their LA showroom but the speakers weren't working. It's funny that I've cloned it's lenses in 20 different permutations but I haven't heard the Beolab 5 in over ten years, way back when B&O had a store in Portland, which has since closed.
As for WAF, I think the exciting thing about the CBT is that you can basically 'dial in' how 'pinpoint' the image is. From what I can see with my experiments, the more you shade the array, the more 'pinpoint' it gets.
That's incorrect. Your wrongly mixing phase issues and output. Whereas the shading removes a good amount of the added dB in the highs you normally get with multiple drivers, there's still an increase in output. The sensitivity of the tweeter in CBT36 is extremely low, in the area of 70-73 dB (can't remember exactly). So there is actually an increase of 6-9 dB in the highest frequencies.
According to the manual, the efficiency at 8khz is 79dB. That's consistent with what I am arguing here:
The sensitivity of a line array is no greater than a single element when the wavelengths are shorter than the elements.
Dayton is kinda hiding that fact by not quoting a sensitivity figure above 8khz. Basically 8khz is right in the range where this phenomenon is occurring, it's the frequency where the comb filtering is beginning for the tweeters in the CBT 36. They're 12.5mm in diameter; that's 27.2khz. We'd expect to see comb filtering begin right around a third of that number, or 9khz, which is exactly why Dayton is quoting a sensitivity rating at EIGHT khz.
http://www.daytonaudio.com/media/resources/301-980-epique-cbt36-assembly-manual.pdf
"14.1.12.4. Sensitivity (2.83 Vrms/1m):
Note: the raw sensitivity (no crossover or EQ) of the CBT36 is frequency dependent. It is roughly flat from 80 to 300
Hz and then rolls off at 3 dB/octave (10 dB/decade) up to 20 kHz. See later section “CBT36 Power Rolloff” in
Appendix 2 for further explanation of this rolloff. For more details see Fig. 26 in this section. Here are some
approximate sensitivity numbers at different frequencies:
80 to 300 Hz: 94 dB
800 Hz: 89 dB
8 kHz: 79 dB"
Last edited:
Nope. The sensitivity of the driver is lower than 79dB. It' closer to 70 dB, thus there's is an increase compared to using a single driver. The tweeter in CBT36 is both very small and a full range driver, hence the extremely low sensitivity. With appropriate drivers one can achieve very high sensitivity in a CBT speaker.
You're mixing phase issues (lobing and combing) with output. Two different things. Both lobing and output is increased with multiple drivers. It's the shading in the CBT that minimizes the output.
@Patrick Bateman, did you follow the Infinite Line Source Theory thread?
It will show the 3 dB/octave drop in SPL output, even for a theoretical ideal model.
http://www.diyaudio.com/forums/multi-way/314917-infinite-line-source-analysis-14.html#post5301172
This has nothing to do with the suspected starting point of combing, which you simply cant predict as being a third of the size of the tweeters.
It will show the 3 dB/octave drop in SPL output, even for a theoretical ideal model.
http://www.diyaudio.com/forums/multi-way/314917-infinite-line-source-analysis-14.html#post5301172
This has nothing to do with the suspected starting point of combing, which you simply cant predict as being a third of the size of the tweeters.