• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

DIY desktop speakers the ASR way - help needed!

OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
Hey @andreasmaaan, thanks for the input!

I have a pair of Don Keele's CBT24 speakers, and they definitely don't appear to have a sound source below the floor. In his 2005 paper (Keele et al 2005) we see the following sound pressure simulations for a legendre shaded curved array at 500Hz and 2KHz. You can see the even field extends above the top of the array at 1m distance.

You are right physical curving for delay is definitely simpler in some senses, but for short arrays the degree of curvature may get rather extreme. I haven't done the calculations yet but Rick Craig's experimental ones had almost a 90 degree curvature and required quite tricky woodworking for the cabinets (See PE Tech Talk thread Post 59). For initial experimentation at least a flat array and DSP would be more flexible and according to Don's research has only minor downsides for the soundfield

How the use of BMRs would affect this I have no serious idea, but in principle the wider the directivity of the drive units and the smaller they are the better, as the simulations are based on ideal transducers spaced only 1cm apart. In practice, as shown by the CBT24s, very good results can be achieved with much less than ideal designs. XO would likely be around 200Hz.

Screen Shot 2021-01-16 at 5.47.44 pm.png
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I have a pair of Don Keele's CBT24 speakers, and they definitely don't appear to have a sound source below the floor.

Yes, in hindsight I articulated myself poorly. The apparent source should be the centre of the arc created by the physical array and its reflection on the floor/desk. In practice, this will place it at the level of the floor/desk, some distance (the radius length of the arc) behind the speakers.

Is this your experience with the CBT24?

You are right physical curving for delay is definitely simpler in some senses, but for short arrays the degree of curvature may get rather extreme.

It depends how wide/narrow your want the vertical directivity to be. For narrower dispersion, the curvature need not be too extreme (I haven't modelled it for your specific circumstances though). And your array will be almost 3 times the height of that one in the link, which will require lesser curvature for a given vertical beamwidth.

How the use of BMRs would affect this I have no serious idea, but in principle the wider the directivity of the drive units and the smaller they are the better, as the simulations are based on ideal transducers spaced only 1cm apart. In practice, as shown by the CBT24s, very good results can be achieved with much less than ideal designs. XO would likely be around 200Hz.

Yeh, on paper the BMRs are certainly excellent candidates for this. I would just measure them first and then model the whole speaker to make sure they will perform as well in practice as they should in theory :)
 
Last edited:
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
Yes, in hindsight I articulated myself poorly. The apparent source should be the centre of the arc created by the physical array and it's reflection on the floor/desk. In practice, this will place it at the level of the floor/desk, some distance behind the speakers.

Is this your experience with the CBT24?

No, not really. The apparent source is towards the top of the speakers themselves, but honestly that might just be visual anchoring. The sound field is so even that 'height' isn't even a property I would really give it. There are also no significant ceiling reflections, so again the sound is rather different to a standard speaker. When you go to an outdoor rock concert does the sound appear to come from those big curved arrays above the stage, or just from the stage?
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
No, not really. The apparent source is towards the top of the speakers themselves, but honestly that might just be visual anchoring. The sound field is so even that 'height' isn't even a property I would really give it. There are also no significant ceiling reflections, so again the sound is rather different to a standard speaker. When you go to an outdoor rock concert does the sound appear to come from those big curved arrays above the stage, or just from the stage?

Well, it depends on the type of array. And of course, visual cues tend to override auditory ones.

In a technical sense, though, the waveform produced by a CBT is equivalent to that of a point source located at the centre of the circle formed by the CBT's arc (including, in the case of a CBT that utilises the floor reflection, the arc formed by the physical array and its reflection).

When I have a bit more time I will demonstrate this mathematically/geometrically. But I'd need to go back through Keele's papers to do this.

Note that the apparent source would indeed be (behind) the top (or just below the top) of the CBT for a listener positioned up high (relative to the top of the CBT), where the CBT has a shallow curvature, and where the listener is a significant distance away (MS Paint diagram obviously not to scale!):

1610794100268.png
 
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
In a technical sense, though, the waveform produced by a CBT is equivalent to that of a point source located at the centre of the circle formed by the CBT's arc (including, in the case of a CBT that utilises the floor reflection, the arc formed by the physical array and its reflection).

What you say would be true of an unshaded curved array, and therefore also of a similar half-array on an acoustically reflective surface. See Figure 45 in the 2005 paper for a diagram, and figures 48-55 for the simulated soundfields. Here's the 4KHz field, which looks I think quite a lot like what you have drawn in your example:

Screen Shot 2021-01-16 at 7.31.17 pm.png


The soundfields in my post above (figures 84 and 86 from the 2005 paper) are with the shading applied and look very different to the fields from a standard curved array. That's what makes them so remarkable! This makes for a quite different listening experience compared with traditional speakers.

Screen Shot 2021-01-16 at 5.47.44 pm.png
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
What you say would be true of an unshaded curved array, and therefore also of a similar half-array on an acoustically reflective surface. See Figure 45 in the 2005 paper for a diagram, and figures 48-55 for the simulated soundfields. Here's the 4KHz field, which looks I think quite a lot like what you have drawn in your example:

View attachment 106437

The soundfields in my post above (figures 84 and 86 from the 2005 paper) are with the shading applied and look very different to the fields from a standard curved array. That's what makes them so remarkable! This makes for a quite different listening experience compared with traditional speakers.

Those diagrams don't tell us the virtual source of the sound, but rather the pressure magnitude distribution. And even if they did, they would place it on the floor in front of the CBT for distances greater than c. 2m (see approx. tangents to the curve following the shape of the distribution, which I've drawn on the 2kHz diagram):

1610798017531.png


But anyway, this is really a side issue, so let's not get bogged down in it :)
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,520
Likes
7,033
Location
Stow, Ohio USA
While I get your requirements, have already been down a comparable path and abandoned. Here is why:
  1. Approached @Selah Audio and he was not satisfied with his initial desktop CBT results. Since my fronts are CBT24s, wanted CBT surrounds. He had another design was more appropriate, but NLA and pricey.
  2. Had used my Mirage OMD-5s for desktop and generally liked them, but they are now on surround duty...
  3. Replaced the Mirages with a stereo pair of Amazon Echo speakers and really like. I doubt they are the sound quality you are targeting, but you might get that from Sonos or Apple.
This said, am really interested in your attempt at a small CBT. But from experience, I suspect you may have better luck with buying or building some small Omnis. Hope this helps. :cool:
 
Last edited:
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
@mcdn just for a very quick answer to this question from Keele himself, see here from about 5m onwards.

Aah, I think we are using the same words to mean different things :) Both views are true I think. The apparent source of the sound waves is a point near the floor, but that's not where the sound appears to be coming from when you listen.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Aah, I think we are using the same words to mean different things :) Both views are true I think. The apparent source of the sound waves is a point near the floor, but that's not where the sound appears to be coming from when you listen.

Fair compromise :)

Having reflected further on your comment that the CBT24's sound appears to come from the top of the speaker (which, given their height of 60", is presumaby not located on the line connecting a seated listener to the midpoint of the centre of curvature), I wonder if you have some absorptive covering (carpet, etc) on the floor between you and the speaker? This would prevent higher-frequency floor reflections, which would drag the apparent source upwards to some position above the connecting line.
 
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
Fair compromise :)

Having reflected further on your comment that the CBT24's sound appears to come from the top of the speaker (which, given their height of 60", is presumaby not located on the line connecting a seated listener to the midpoint of the centre of curvature), I wonder if you have some absorptive covering (carpet, etc) on the floor between you and the speaker? This would prevent higher-frequency floor reflections, which would drag the apparent source upwards to some position above the connecting line.

They're on a solid brick floor so I don't think that's it. I'll have to reread the paper but I'm pretty sure that slide is a bit of a simplification. With shading the apparent wave source is much much farther back than without shading. As I understand it that's the 'constant' part of a CBT, that the beam width (i.e. height in this application) doesn't vary with distance.

Maybe we should break out a new thread for the theory side of this?
 

Mashcky

Active Member
Forum Donor
Joined
Feb 19, 2019
Messages
121
Likes
144
Location
Burlington, Vermont
The full-range with woofer is a good idea because it solves the polar response issue of being close to your speakers without using coaxial while saving some baffle width if you side-mount a woofer. Yet another option suited for this configuration would be the ScanSpeak 10f/8414.
 
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
Thanks for the feedback and suggestions everyone. I've decied to do a bit of a design update based on it all

For the array drivers I've decided to switch to the smaller TEBM35C10-4. It's well reviewed by Erin, and smaller so the units can be more tightly spaced. Also switching to a single Dayton LW150 as a woofer. The Tang-Band stuck out a lot from the baffle and is deeper than the Dayton as well.

This means the box can now be just 55mm wide, with a slightly reduced height of 420mm using 8 drivers per array spaced at 52mm. The tradeoff will be reduced output in the bass - 89dB @ 50Hz probably. Crossover point might as well move to around 300Hz, as the array loses soundfield control below that anyway because it's quite short.

For experimentation I'm going to use a cheap USB 8 channel soundcard and do all the DSP on the host computer. The bottom two drive units can probably be treated as one, sparing one channel for the woofer. Now I just need 9 channels of cheap amplification per side! Anybody got some genius AliExpress links to share? Roughly 40W @ 1% THD needed for the woofer, 8 x 10W @ 1% THD needed for the BMRs.

Picture with dimensions, please imagine the BMRs on the front :)

Screen Shot 2021-01-17 at 11.35.30 am.png
 

Selah Audio

Active Member
Audio Company
Forum Donor
Joined
Feb 4, 2020
Messages
194
Likes
445
While I get your requirements, have already been down a comparable path and abandoned. Here is why:
  1. Approached @Selah Audio and he was not satisfied with his desktop CBT results. Since my fronts are CBT24s, wanted CBT surrounds. He had another design was more appropriate, but NLA and pricey.
  2. Had used my Mirage OMD-5s for desktop and generally liked them, but they are now on surround duty...
  3. Replaced the Mirages with a stereo pair of Amazon Echo speakers and really like. I doubt they are the sound quality you are targeting, but you might get that from Sonos or Apple.
This said, am really interested in your attempt at a small CBT. But from experience, I suspect you may have better luck with buying or building some small Omnis. Hope this helps. :cool:

Here are two desktop CBT arrays I have designed. The smaller one was with four 2.5" drivers. This was a proof-of-concept array which worked well except for the limitations of a small "full-range" driver. This prompted the 2-way array which is powered with a DSP amplifier and is larger but fits fine on my desk.
 

Attachments

  • dB Desktop Array.JPG
    dB Desktop Array.JPG
    181.6 KB · Views: 151
  • Desktop1.JPG
    Desktop1.JPG
    186.5 KB · Views: 149
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
Here are two desktop CBT arrays I have designed. The smaller one was with four 2.5" drivers. This was a proof-of-concept array which worked well except for the limitations of a small "full-range" driver. This prompted the 2-way array which is powered with a DSP amplifier and is larger but fits fine on my desk.

Neat! Do you have any tips or learnings?
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,520
Likes
7,033
Location
Stow, Ohio USA
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
Ooh, those look interesting, thanks Rick! I'm hoping to have a prototype of mine to share in a month or so.
 

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,520
Likes
7,033
Location
Stow, Ohio USA
Ooh, those look interesting, thanks Rick! I'm hoping to have a prototype of mine to share in a month or so.

I will be interested to see how well they review. They seem a bit pricey. I did not pay much more than that for CBT24s!
 
OP
M

mcdn

Addicted to Fun and Learning
Forum Donor
Joined
Mar 7, 2020
Messages
559
Likes
780
I will be interested to see how well they review. They seem a bit pricey. I did not pay much more than that for CBT24s!

Finished speaker pricing is terribly sensitive to production volume it seems. Parts Express I would imagine has the volumes and marketing reach to sell almost anything at a keen price. A new boutique entrant has a bit of a dilemma - do you price to make each initial sale profitable, or lower the price and hope to make it up in volume? This isn't about the BOM (usually) but all the other costs.
 
Top Bottom