I am wondering how best to design a speaker to have the tightest lobe, beam, or dispersion. Usually, we try to have dispersion be as wide as possible, so the speaker sweet spot is wide. Here let’s go in the opposite direction.
One application might be to make a speaker for Dolby Atmos in which you can bounce sound off a ceiling. I know that such reflections are not a good way to go for Atmos. But I can’t reach my ceiling easily and don’t want speakers up there. (15 feet high, vaulted). Mostly, though, this could be a fun design exercise and I’m looking for a next speaker project. As simple as it may seem, I have not been able to find enough information on this topic. So let’s see what would make for the best design.
What is out there already:
I’m probably missing a lot. But here is what I have found so far. Most Atmos speakers look like the following. Basically, an MT speaker angled up. Probably spraying sound all over the place. Not good. Let’s not do that.
There is an interesting thread here where horns and parabolic reflectors are used. But the conversation moves toward needing stadium level projection distances and the designs don’t seem suitable for, say, a house that you actually live in. 😀
https://www.diyaudio.com/community/threads/extremely-narrow-directivity-speaker-design.358588/
There are ultrasonic arrays, which give very narrow dispersion. My limited understanding is that they are not really good for music or movies.
Then there are speakers with 4 drivers arranged in a square. Here is where things start to get really interesting. At least to me. There is a commercial option from Triad. And one DIY option:
https://www.avsforum.com/threads/javs-diy-dolby-atmos-mkii-modules.2378314/
Four drivers in a square:
Let’s explore this idea some more. Will it work well? I don’t know. Any thoughts out there?
Look at the classic MTM speaker design. Note how the middle, dashed plot (2nd order Linkwitz-Riley, LR2) shows a nice, single lobe. Looks great so far. This image is for vertical directivity. If we also want narrow horizontal directivity, we would have a horizontal MTM. (Like some center channels.) Basically, we end up with the four drivers in a square plus a tweeter in the middle.
Crossover or no?
In the image above, notice how widely different the dispersion is depending on the crossover choice. An LR2 appears to be best for this current purpose. I do not know if that LR2 is electric, acoustic, or both. At any rate, I know that hitting LR2 is very challenging when designing a speakers. I have a couple of designs under my belt and was able to hammer things into shape and hit LR4. But LR2 is tough. And if we stray away from LR2, according to the image above, we may quickly lose the whole point of the exercise here.
So maybe it’s best to have no crossover at all? A full range speaker. With 4 drivers in a square. Like the Triad and DIY designs mentioned above?
Which drivers:
I guess the first question here is whether or not there will be a tweeter used. If there will be no tweeter, I guess just find the overall best performing full range driver. If we have a tweeter and a crossover, maybe try to find ones that are as flat as possible in the crossover region to have some hope of hitting an LR2.
Another option might be to have both a tweeter and woofers, but try to, effectively, have no crossover. Find a tweeter with relatively flat frequency response. And a woofer that is also flat. And for each, they have a natural tendency to drop off output around the same frequencies. So that they, effectively, sum up to be about flat without any crossover. I tried a project like this for an MT design once and it came out OK, not super. Add a resistor to drop the tweeter output down to the level of the woofers. Add a low pass cap to the tweeter to prevent it from getting fried from pops when you turn on the amp. Otherwise no real crossover to alter the directivity.
As frequencies go low, directivity also decreases. So there is probably not much need for the drivers to play below, say, 100 Hz. But we can always cut that off with the processor, anyways.
Ideas?
I am curious to see what thoughts everyone might have on this topic. Thanks!
One application might be to make a speaker for Dolby Atmos in which you can bounce sound off a ceiling. I know that such reflections are not a good way to go for Atmos. But I can’t reach my ceiling easily and don’t want speakers up there. (15 feet high, vaulted). Mostly, though, this could be a fun design exercise and I’m looking for a next speaker project. As simple as it may seem, I have not been able to find enough information on this topic. So let’s see what would make for the best design.
What is out there already:
I’m probably missing a lot. But here is what I have found so far. Most Atmos speakers look like the following. Basically, an MT speaker angled up. Probably spraying sound all over the place. Not good. Let’s not do that.
There is an interesting thread here where horns and parabolic reflectors are used. But the conversation moves toward needing stadium level projection distances and the designs don’t seem suitable for, say, a house that you actually live in. 😀
https://www.diyaudio.com/community/threads/extremely-narrow-directivity-speaker-design.358588/
There are ultrasonic arrays, which give very narrow dispersion. My limited understanding is that they are not really good for music or movies.
Then there are speakers with 4 drivers arranged in a square. Here is where things start to get really interesting. At least to me. There is a commercial option from Triad. And one DIY option:
https://www.avsforum.com/threads/javs-diy-dolby-atmos-mkii-modules.2378314/
Four drivers in a square:
Let’s explore this idea some more. Will it work well? I don’t know. Any thoughts out there?
Look at the classic MTM speaker design. Note how the middle, dashed plot (2nd order Linkwitz-Riley, LR2) shows a nice, single lobe. Looks great so far. This image is for vertical directivity. If we also want narrow horizontal directivity, we would have a horizontal MTM. (Like some center channels.) Basically, we end up with the four drivers in a square plus a tweeter in the middle.
Crossover or no?
In the image above, notice how widely different the dispersion is depending on the crossover choice. An LR2 appears to be best for this current purpose. I do not know if that LR2 is electric, acoustic, or both. At any rate, I know that hitting LR2 is very challenging when designing a speakers. I have a couple of designs under my belt and was able to hammer things into shape and hit LR4. But LR2 is tough. And if we stray away from LR2, according to the image above, we may quickly lose the whole point of the exercise here.
So maybe it’s best to have no crossover at all? A full range speaker. With 4 drivers in a square. Like the Triad and DIY designs mentioned above?
Which drivers:
I guess the first question here is whether or not there will be a tweeter used. If there will be no tweeter, I guess just find the overall best performing full range driver. If we have a tweeter and a crossover, maybe try to find ones that are as flat as possible in the crossover region to have some hope of hitting an LR2.
Another option might be to have both a tweeter and woofers, but try to, effectively, have no crossover. Find a tweeter with relatively flat frequency response. And a woofer that is also flat. And for each, they have a natural tendency to drop off output around the same frequencies. So that they, effectively, sum up to be about flat without any crossover. I tried a project like this for an MT design once and it came out OK, not super. Add a resistor to drop the tweeter output down to the level of the woofers. Add a low pass cap to the tweeter to prevent it from getting fried from pops when you turn on the amp. Otherwise no real crossover to alter the directivity.
As frequencies go low, directivity also decreases. So there is probably not much need for the drivers to play below, say, 100 Hz. But we can always cut that off with the processor, anyways.
Ideas?
I am curious to see what thoughts everyone might have on this topic. Thanks!
The approach will be highly dependent on what frequency range you want it to operate over, how directional you really want it, and exactly how the drivers behave. In the upper midrange/treble, an array of small full-range drivers typically gets quite directional at higher frequencies, but this is much higher than your suggested 100 Hz. A cardioid array might work better if you need to go low. If you want to build a driver, a large square planar speaker that's playing midrange and up also gets quite directional.
In your 4 drivers in a square array, if the individual speakers have been optimized for broader dispersion at higher frequencies, they will start to behave more like individual spaced point sources higher in frequency, which will limit the directivity of the array.
Four 4-inch array, assuming pistonic behavior
Four 1-inch drivers with same center spacing
On the large, highly processed array front, there are things like Holoplot. Smaller soundbars have also used DSP to do beamforming, but not to this degree.
https://audioxpress.com/news/holopl...amforming-and-wavefield-synthesis-at-ise-2024
https://audioxpress.com/news/audios...d-beamforming-technology-and-head-tracking-ai
"Audioscenic will finally see its 3D audio beamforming technology reach consumers in a product from Razer, one of the world’s leading lifestyle brands for gamers. The Razer Leviathan V2 Pro, is the world’s first desktop soundbar with THX Spatial Audio combined with Audioscenic unique technology that allows a truly personalized, immersive listening experience without the need for headphones."
https://audioxpress.com/article/r-d-stories-audioscenic-from-vision-to-product
In your 4 drivers in a square array, if the individual speakers have been optimized for broader dispersion at higher frequencies, they will start to behave more like individual spaced point sources higher in frequency, which will limit the directivity of the array.
Four 4-inch array, assuming pistonic behavior
Four 1-inch drivers with same center spacing
On the large, highly processed array front, there are things like Holoplot. Smaller soundbars have also used DSP to do beamforming, but not to this degree.
https://audioxpress.com/news/holopl...amforming-and-wavefield-synthesis-at-ise-2024
https://audioxpress.com/news/audios...d-beamforming-technology-and-head-tracking-ai
"Audioscenic will finally see its 3D audio beamforming technology reach consumers in a product from Razer, one of the world’s leading lifestyle brands for gamers. The Razer Leviathan V2 Pro, is the world’s first desktop soundbar with THX Spatial Audio combined with Audioscenic unique technology that allows a truly personalized, immersive listening experience without the need for headphones."
https://audioxpress.com/article/r-d-stories-audioscenic-from-vision-to-product
Last edited:
The Yamaha YSP-1 Digital Sound Projector was another interesting one
https://www.soundandvision.com/content/yamaha-ysp-1-digital-sound-projector
"The YSP-1 is a single speaker cabinet that reflects sound beams off the walls to create the aural illusion of a five-speaker setup. It uses two 4.375-inch woofers and an array of 40 1.375-inch tweeters to accomplish the feat. With the push of a button, you can set these imaginary speakers to emulate four different surround setups: stereo, three beam (LCR), five beam, and stereo plus three"
https://www.soundandvision.com/content/yamaha-ysp-1-digital-sound-projector-page-2
https://www.soundandvision.com/content/yamaha-ysp-1-digital-sound-projector
"The YSP-1 is a single speaker cabinet that reflects sound beams off the walls to create the aural illusion of a five-speaker setup. It uses two 4.375-inch woofers and an array of 40 1.375-inch tweeters to accomplish the feat. With the push of a button, you can set these imaginary speakers to emulate four different surround setups: stereo, three beam (LCR), five beam, and stereo plus three"
https://www.soundandvision.com/content/yamaha-ysp-1-digital-sound-projector-page-2
Hi mattstat,
Thanks for the thoughts. Those are interesting products. So maybe the idea of many drivers in a 2D array is a viable way to go? That Holoplot has many drivers in a 2D array. Although not exactly lined up perfectly. Maybe for a reason. I wonder what sort of crossover or electronic shaping is in there. But maybe the idea here of 4 drivers in a square is a start in this direction. Hard to know, at least for me.
And thanks for those simulations. Presumably the general drop off of the 4 inch drivers at high frequencies (>1,000 Hz) is just because they are not tweeters? Those nulls, however, are pretty extreme, -20 dB and such. Hmmm. Maybe from the drivers cancelling each other out at certain frequencies?
Taking this info and some far leaps… Maybe a small 2D array of drivers would be good. But then they need to be spaced apart so that they all do not cancel each other out at the same frequencies. Kind of like an offset square. Like in the right image of the Holoplot. The drivers are not aligned all the same. The ideal could be that for any of the 4 drivers, the distance to the other 3 are 3 different distances. This could be similar to how tweeters are often offset from the center of the baffle. (It’s square root of 2 from the crossover frequency or something like that.)
So place 4 drivers together in something resembling a square. Offset them each from the others. Hmmm. Trying to think through what would work here.
Thanks for the thoughts. Those are interesting products. So maybe the idea of many drivers in a 2D array is a viable way to go? That Holoplot has many drivers in a 2D array. Although not exactly lined up perfectly. Maybe for a reason. I wonder what sort of crossover or electronic shaping is in there. But maybe the idea here of 4 drivers in a square is a start in this direction. Hard to know, at least for me.
And thanks for those simulations. Presumably the general drop off of the 4 inch drivers at high frequencies (>1,000 Hz) is just because they are not tweeters? Those nulls, however, are pretty extreme, -20 dB and such. Hmmm. Maybe from the drivers cancelling each other out at certain frequencies?
Taking this info and some far leaps… Maybe a small 2D array of drivers would be good. But then they need to be spaced apart so that they all do not cancel each other out at the same frequencies. Kind of like an offset square. Like in the right image of the Holoplot. The drivers are not aligned all the same. The ideal could be that for any of the 4 drivers, the distance to the other 3 are 3 different distances. This could be similar to how tweeters are often offset from the center of the baffle. (It’s square root of 2 from the crossover frequency or something like that.)
So place 4 drivers together in something resembling a square. Offset them each from the others. Hmmm. Trying to think through what would work here.
No. Those were ideal drivers playing full range in the simulation. All of the fall-off at high frequencies was related to array and driver directionality. On axis, they look like this:
Yes. As you go off axis, the different path lengths start to cause constructive/destructive interference. It's also worth noting that larger drivers have similar issues in a simulation like this. For an 8-inch driver with flat on-axis response, it looks like this off axis:
And again, these are ideal drivers with perfect pistonic behavior. In real life, speakers have breakup, roll-off, and some are designed to intentionally decouple the center from the edge at higher frequencies to improve dispersion. In reality, things aren't likely to be as extreme as these models show. But arrays of small drivers are more likely to behave as simplified models predict.
To me, the cancellation was the desired property, since you want a directional array. As you space the drivers out, you progressively lose some of the cancellation, but also directivity. Here's the same array with the spacing doubled:
The Holoplot stuff is extreme and just an example of what can be done with large arrays and lots of processing. They seem to be running custom hardware and software, so not something the average person can do.
https://audioxpress.com/article/r-d-stories-design-and-development-of-the-holoplot-x1-matrix-array
"The Matrix Array is the logical continuation of two already existing landmark innovations in pro audio — the line array and electronic beamsteering — and provides control of sound in both horizontal and vertical axes through a matrix of loudspeaker drivers.
. . .
The X1 product series features two audio modules. The Modul 96 (MD96) is a two-way audio module integrating 96 drivers in a two-layered matrix design. . . . X1 offers a wide range of previously inaccessible sound control capabilities. Each module can create up to 12 individual and independent beams, able to reproduce differing content at the same time. With just one centralized Matrix Array, sound propagation can be tailored to cover an audience area and precisely target one or multiple zones.
. . .
Helmut came up with what he named “holophony.” His holophonic approach involves the use of Wave Field Synthesis and two-dimensional speaker arrays, which was the foundation for his first patent back in 2004. Out of this patent, Holoplot was born to turn that theoretical vision into reality.
. . .
Holoplot’s approach to Wave Field Synthesis was different from the traditional approach — most had combined signal processing with traditional point source loudspeakers, and required the installation of those loudspeakers all around the listener to create an immersive effect. Holoplot’s proprietary loudspeaker and processing technology allowed us to handle the wave propagation in a completely new and more finely controlled way. We were for example able to create virtual sound sources through room reflections — meaning it was for the first time possible to expand traditional WFS and create a fully immersive audio experience utilizing just a single Matrix Array.”
. . .
Creating X1 certainly hasn’t been a simple process, or a natural one. The issues Holoplot faced in bringing this technology to life have been many. For example, X1 uses an unheard of number of drivers, each one individually driven and signal processed via 8700 digitally controlled filters computed in real time.
The amount of amplifier channels required to achieve this inevitably leads to a huge amount of heat to manage, and a seemingly insurmountable amount of DSP needs to be computed in real-time. To put this into perspective, a relatively small 3×3 X1 Matrix Array usually has more amplifier channels than the largest installation of conventional loudspeaker systems.
. . .
“The biggest challenges we face when developing a beamsteered product are the array size and driver spacing. At low frequencies we need a large array to obtain sufficient directivity control. At high frequencies a small array will do, but a very dense driver distribution is required in order to avoid spatial aliasing. Meeting all of these physical requirements is practically very challenging,”
. . .
X1 is software-driven hardware. Both elements are intrinsically linked and interdependent."
Last edited:
Ahhhh, OK. Those simulations are the off axis responses. And on axis is flat. Now it’s easier to see what you are saying. 😀
Looking at the options, it seems like the 4 drivers close together would be a little better than the 4 drivers farther apart. Generally less output for the closer spacing above, say, 1,000 Hz.
But the single 8” driver might be even better than the 4 closely spaced ones. I would imagine that a single 4” driver would look similar to the 8” in the simulation, just with differences in the frequencies of the nulls from edge effects.
So the drop-off at higher frequencies when off axis is good. The peaks and valleys maybe not. Well, better to have the nulls than not. But maybe the whole plot could still have the downward slope with increasing frequencies, but be smoothed out by arranging the 4 drivers such that the distances between all 4 are different. You would still get cancellations, but maybe at different frequencies. And the plot would be smoother. The nulls would not be as deep. But the peaks would be lower. Am I thinking about that correctly?
It’s interesting to see the single 8” do so well. So maybe there is no need for an array of 4 drivers? (An 8" would definitely need a tweeter and it's not yet clear to me what the crossover implications would be.)
Dealing with gaining more directivity/less dispersion below 1,000 Hz is a separate matter. I’ll worry about that after I can get my head around what is happening above 1,000 Hz. I realize that once we’re down to frequencies with longer wavelengths (say 200 Hz) there may be nothing to do. Crossover or not is another issue to grapple with when I have some better understanding of the best driver arrangement.
For the Holoplot I would think that, with enough processing power, you might be able to really tune which drivers are playing which frequency when and have everything cancel when you want. Getting that to all work out would be more of a career than a hobbyist speaker project. But cool. Maybe a small array here can get into the ballpark. I'm still not sure.
Looking at the options, it seems like the 4 drivers close together would be a little better than the 4 drivers farther apart. Generally less output for the closer spacing above, say, 1,000 Hz.
But the single 8” driver might be even better than the 4 closely spaced ones. I would imagine that a single 4” driver would look similar to the 8” in the simulation, just with differences in the frequencies of the nulls from edge effects.
So the drop-off at higher frequencies when off axis is good. The peaks and valleys maybe not. Well, better to have the nulls than not. But maybe the whole plot could still have the downward slope with increasing frequencies, but be smoothed out by arranging the 4 drivers such that the distances between all 4 are different. You would still get cancellations, but maybe at different frequencies. And the plot would be smoother. The nulls would not be as deep. But the peaks would be lower. Am I thinking about that correctly?
It’s interesting to see the single 8” do so well. So maybe there is no need for an array of 4 drivers? (An 8" would definitely need a tweeter and it's not yet clear to me what the crossover implications would be.)
Dealing with gaining more directivity/less dispersion below 1,000 Hz is a separate matter. I’ll worry about that after I can get my head around what is happening above 1,000 Hz. I realize that once we’re down to frequencies with longer wavelengths (say 200 Hz) there may be nothing to do. Crossover or not is another issue to grapple with when I have some better understanding of the best driver arrangement.
For the Holoplot I would think that, with enough processing power, you might be able to really tune which drivers are playing which frequency when and have everything cancel when you want. Getting that to all work out would be more of a career than a hobbyist speaker project. But cool. Maybe a small array here can get into the ballpark. I'm still not sure.
The issue with asymmetric arrays is that off-axis response is going to be different in every direction. You can come up with a pattern that works really well in one direction, but it typically won't hold for all axes.
Adding a single regular tweeter will create broad dispersion over most of its range, since its size is so small. This behavior is covered by the general rule that drivers start getting directional when their diameter is approximately equal to 1/3 - 1/2 wavelength (not a firm rule - it's all on a continuum). A simple array of tightly spaced small drivers is basically working with the same phenomenon - making the array large enough that directivity increases.
Using a horn tweeter will restrict dispersion, but if you want it to hold a narrow pattern down low, it's typically going to be larger. Waveguides also work.
Finding an off-the-shelf 8-inch full-range highly directional driver is going to be tough. Most full-ranges are designed not to behave that way. Many drivers include off-axis response in their plots. A couple examples:
Seas Prestige FA22RCZ (H1597-08) 8" Fullrange
https://www.madisoundspeakerstore.com/approx-8-fullrange/seas-prestige-fa22rcz-h1597-08-8-fullrange/
SB Acoustics SB20FRPC30-8 8" Paper Full Range
https://www.madisoundspeakerstore.c...sb-acoustics-sb20frpc30-8-8-paper-full-range/
And a small horn loaded tweeter vs plain (same dome tweeter is in both)
https://www.madisoundspeakerstore.c...2606/9200-horn-loaded-1-textile-dome-tweeter/
https://www.madisoundspeakerstore.c...-discovery-d2606/9200-1-textile-dome-tweeter/
Adding a single regular tweeter will create broad dispersion over most of its range, since its size is so small. This behavior is covered by the general rule that drivers start getting directional when their diameter is approximately equal to 1/3 - 1/2 wavelength (not a firm rule - it's all on a continuum). A simple array of tightly spaced small drivers is basically working with the same phenomenon - making the array large enough that directivity increases.
Using a horn tweeter will restrict dispersion, but if you want it to hold a narrow pattern down low, it's typically going to be larger. Waveguides also work.
Finding an off-the-shelf 8-inch full-range highly directional driver is going to be tough. Most full-ranges are designed not to behave that way. Many drivers include off-axis response in their plots. A couple examples:
Seas Prestige FA22RCZ (H1597-08) 8" Fullrange
https://www.madisoundspeakerstore.com/approx-8-fullrange/seas-prestige-fa22rcz-h1597-08-8-fullrange/
SB Acoustics SB20FRPC30-8 8" Paper Full Range
https://www.madisoundspeakerstore.c...sb-acoustics-sb20frpc30-8-8-paper-full-range/
And a small horn loaded tweeter vs plain (same dome tweeter is in both)
https://www.madisoundspeakerstore.c...2606/9200-horn-loaded-1-textile-dome-tweeter/
https://www.madisoundspeakerstore.c...-discovery-d2606/9200-1-textile-dome-tweeter/
Last edited:
Your problem is the ceiling height at 15ft……the total distance to the listener is far too great creating too much delay for object based processing to accomplish.
2 years back I was called in by a friends firm for a solution to a similar problem with a cathedral ceiling. The only viable solution was hanging pendant ATMOS speakers which the client rejected. A 9.2 system was used instead and did a fantastic job without object based processing which IMO is too limited in its application to be viable at the expected performance levels of the algorithm. Take a minute to read and research how OB works and you’ll quickly understand my last statement…..the juice ain’t worth the squeeze .
2 years back I was called in by a friends firm for a solution to a similar problem with a cathedral ceiling. The only viable solution was hanging pendant ATMOS speakers which the client rejected. A 9.2 system was used instead and did a fantastic job without object based processing which IMO is too limited in its application to be viable at the expected performance levels of the algorithm. Take a minute to read and research how OB works and you’ll quickly understand my last statement…..the juice ain’t worth the squeeze .
I understand completely. I’ve read how ceiling reflection Atmos speakers are very poor. However maybe there is a way to make something better. More relevant is that I’ve now filled up the house and office with all of the nice speakers that can be used. Even my gym is now 2.1. So I’m up for something different to try. Maybe a somewhat unconventional design. It’s a hobby. If it works only a little, that’s fine. 😀
This comment is particularly insightful. Thanks! So a tweeter (conventional or horn, etc.) would be bad here because the small diameter means that it will only be directional at the very highest frequencies. We’d want the largest possible driver. But we also want this mythical driver to be full range, playing up to ~20,000 Hz so that we do not need a tweeter. The 8” full range drivers have, in effect, tweeters or something of the sort in the middle of the cones, which makes them less directional. Fine for other uses, but not here. That probably puts us into the neighborhood of conventional 3” or 4” full range drivers. Try to find one with the steepest decrease in output when off axis.
Single driver or in an array? A 2x2 square array will have the benefit of the cancellation nulls. But there would also be peaks from constructive interference, I would imagine. If it’s only nulls, then a 3x3 or 4x4 array might be even better?
For basic direct radiators, the constructive addition is the theoretical flat output on-axis at low frequencies, so it's not like you're going to have +6 dB peaks over the low frequency output.
The array size is really a question of what's practical for you and what you want to play with. I generally try to mock things up quickly and cheaply to see how they're going to behave in real life.
I'd probably start by comparing a 2 by 2 array of 4-inch drivers to a 4 by 4 array of 2-inch-ish drivers. A good 2.5 inch driver will play well down to around 300 Hz, which should be low enough to get a feel for the array's behavior.
OK, so arranging drivers to get as many nulls as possible is good. And don’t worry about constructive interference.
If tweeters are too small and not directional enough, then don’t use them and don't worry about a crossover. Go full range. I will guess that a few parts to shape the output of the array of drivers will be OK. Maybe a cap to decrease cone resonance, a Zobel for flatter impedance, maybe an LCR filter here or there, etc. Maybe such things, as needed, will not make the drivers go more broadly dispersed, as can be seen with the MTM topology with different crossovers, shown above?
No major bounds of practicality yet. Just trying to get an idea of what something close to ideal would look like. Then I can chip away from that with practicality- size, cost, impedance issues from wiring a given number of drivers, etc.
Maybe I should do some diffraction simulations first. Like you are doing. That might be enough to decide, in advance, if I want 2x2 or 3x3 or 4x4 arrays. Try each with 2” or 3” or 4” or 5”, etc. drivers. Simulations would help see where the diminishing returns are, where the best nulls are, etc. Then can build real things to test from there.
Looks like you are using Xsim. I’ve only got modern Macs, not PC’s. So under cover of darkness I stole an old PC laptop from someone in the house and downloaded Xsim. Hmmm. Where is the baffle diffraction simulator? I don’t see it, did a quick Google search for info, and can’t find a manual. It’s got to be in there somewhere. I’ll keep looking.
If tweeters are too small and not directional enough, then don’t use them and don't worry about a crossover. Go full range. I will guess that a few parts to shape the output of the array of drivers will be OK. Maybe a cap to decrease cone resonance, a Zobel for flatter impedance, maybe an LCR filter here or there, etc. Maybe such things, as needed, will not make the drivers go more broadly dispersed, as can be seen with the MTM topology with different crossovers, shown above?
No major bounds of practicality yet. Just trying to get an idea of what something close to ideal would look like. Then I can chip away from that with practicality- size, cost, impedance issues from wiring a given number of drivers, etc.
Maybe I should do some diffraction simulations first. Like you are doing. That might be enough to decide, in advance, if I want 2x2 or 3x3 or 4x4 arrays. Try each with 2” or 3” or 4” or 5”, etc. drivers. Simulations would help see where the diminishing returns are, where the best nulls are, etc. Then can build real things to test from there.
Looks like you are using Xsim. I’ve only got modern Macs, not PC’s. So under cover of darkness I stole an old PC laptop from someone in the house and downloaded Xsim. Hmmm. Where is the baffle diffraction simulator? I don’t see it, did a quick Google search for info, and can’t find a manual. It’s got to be in there somewhere. I’ll keep looking.
I think the 3D features have been removed from the newest version. You may need to hunt for the older version if you want those. In XSim3D, the measurement angle features are in the Baffle menu.
There are other free modelers with more features (like full baffle models including diffraction), but they also come with more complexity. XSim is pretty easy to use.
As long as you are applying basic shaping filters to the entire array and it's still playing mostly full-range, the dispersion pattern should be maintained.
Last edited:
That’s an interesting idea. The thought had crossed my mind, initially, but then left once I came across these mini arrays. But dipoles may be worth a look here. Some options that come to mind:
-Just add absorbing material to one side of a dipole. Looks like a 3” thick piece of Owens Corning 703 fiberglass would absorb high frequencies down to maybe around 250 Hz. That could be good for this application. I’ve made absorption panels from OC703 and it’s easy to work with and cut into desired shapes.
-Make the dipole in an MTM configuration with an LR2 crossover (if possible) to take advantage of the beaming in the image in post #1. Add OC703 to the back.
-Make a 2x2 (or 3x3, 4x4, etc.) array of full range drivers on an open baffle and put the OC703 on the back side. So maybe you get the narrow dispersion benefits of both an array plus those of a dipole.
Hmmm…
Ah! OK, so it’s not there and I’m not quite as stupid as I was feeling for not being able to find it. 😀 I’m not finding an older version for download. There are other programs I’ll look into. The Edge looks perfect for this sort of thing, except for it seems to only show on axis, not off axis. AKABAK might be an option as might VituixCAD. There is a diffraction simulator in VituixCAD, which might be good here. I've never tried the program, but I'll play around with it a bit.
I would like to try some modeling of the different options here- 2x2 versus 3x3 arrays, 2” versus 4” drivers, etc. For simple things like this, modeling should be pretty good, enough to get me going to help decide which driver types to focus on and such.
Excellent. I was guessing that simple filters should be fine. However a crossover with two drivers interacting in specific ways might be an issue. I’ve never worked with a full range speaker before. But it makes things easier in some regards. I’ll probably still want to take measurements and shape the output if there is much of a need.
Last edited:
fiberglass is a carcinogen
use Auralex Foam - no need to reinvent the wheel
yes it is expensive but it is also very effective because it is specifically designed for this
i lined the inside of my subwoofer box with auralex foam and when you stick your head in the box where the driver goes ( it's a 18" sub ) you can't even hear yourself scream - completely unnecessary for subs but shows the performance of the material
I’m not sure you do. Object based audio processing relies on an algorithm that uses distance in conjunction with phase to isolate an object sound in a 3D space and move it through the soundstage…..this isn’t an ambient field system. So when the distance from the direct sound to the front mains and the listener is outside of the operational range, the function fails…..the object sound becomes part of the surround/ambient mix. You could create the worlds first extreme directivity speaker and it would not matter for Atmos…..the object is created and steered by the mains and the Atmos speakers through distance and the calculated relation of phase over time.
Thanks for the recommendation. That's interesting. I'm good to spend more money on foam if there are fewer health concerns. If the speaker ends up using foam. Sounds like fun to put your head into the sub cabinet. 😀