I found a very interesting paper on omni directional sound sources.
Paper title is: “An experimental evaluation of regular polyhedron loudspeakers as omni directional sources of sound” from T.M. Leishman e.a.
The paper can be downloaded from: https://scholar.google.nl/scholar?q...sound+pdf&hl=nl&as_sdt=0&as_vis=1&oi=scholart
Tests were done in an anechoic chamber with an circular microphone array on regular polyhedra with 4 to 20 drivers. The sound sources were meant for acoustic room measurements but I think that some results are also valid for sound reproduction applications.
Especially, the “balloon” presentation of the results was an eye-opener for me.
For me, the measurements at 2 and 4 kHz suggest the following:
Paper title is: “An experimental evaluation of regular polyhedron loudspeakers as omni directional sources of sound” from T.M. Leishman e.a.
The paper can be downloaded from: https://scholar.google.nl/scholar?q...sound+pdf&hl=nl&as_sdt=0&as_vis=1&oi=scholart
Tests were done in an anechoic chamber with an circular microphone array on regular polyhedra with 4 to 20 drivers. The sound sources were meant for acoustic room measurements but I think that some results are also valid for sound reproduction applications.
Especially, the “balloon” presentation of the results was an eye-opener for me.
For me, the measurements at 2 and 4 kHz suggest the following:
- The radiation pattern continuously changes when frequency changes. I suppose we don’t notice this because the many room reflections will cancel out these effects.
- The prickly shape of the radiation patterns are caused by diffraction on the polyhedral edges. I assume that the different (truncated) polyhedra will all have different patterns. The size act as a scale factor. These diffraction effects are not always taken into account in theoretical analysis.
- The differences between the different polyhedrons, and with that the number of drivers, are not significant. All are appr. equal good (or bad).
- The diffraction effects cannot be eliminated by EQ methods. At the most, the spl can be make more or less flat in one direction.
It is already established that EQ can do nothing about diffraction.
If you consider that a polyhedron is an approximation of a sphere with flat bits “glued” together. Use smaller panels, and add as much roudnover as you can. A sphere with flat spots only to mount the driver would be the resulkt if you took the concept to the limit. Edge diffraction would then be one driver interfering with another.
Now i’ll go read the paper, you make it seem quite enticing.
dave
I often heard experienced listeners say about omni-directional speakers that the sound was sparkling but that the stereo image seemed a bit woolly.
Some time ago I argued in this topic that diffraction results in a poor spl, leading to a less optimal stereo reproduction for my omni’s.
I now think that it should be possible to greatly reduce the diffraction effects by placing damping material on the outside of the omni’s. By sheer luck, this is possible with the shape I chose for my omni’s.
The omni’s have the shape of a truncated octahedron, consisting of eight hexagons and six squares. The drivers are mounted in four of the hexagons.
Due to this arrangement, there is an empty plane (hexahedron or square) between all the drivers. Damping material can now be placed on these empty surfaces to reduce diffraction.
I took a sheet of melamine foam (4 cm thick) and made the slices from it. This turned out to be quite difficult due to the strange angles and double and triple miter. However, in the end I managed more or less to do so.
Halfway through I realized that only the areas visible from the sweet spot had to be covered. Both speakers were treated in this way.
I didn't bother to take SPL measurements, I was only concerned with the stereo behavior. Because I know of no objective method to measure stereo quality, I just started listening to music.
The result exceeded my expectations. The improvement was immense. The stereo image was sparkling and very detailed. Better than I'd ever heard.
To be honest, I would like to emphasize that I am not an expert in qualitative assessment of loudspeaker systems, nor an experienced stereo listener, but I liked what I heart.
In short, I am now very happy with my omni directional speakers.
My only remaining concern are the WAF aspects.
Jaap
Some time ago I argued in this topic that diffraction results in a poor spl, leading to a less optimal stereo reproduction for my omni’s.
I now think that it should be possible to greatly reduce the diffraction effects by placing damping material on the outside of the omni’s. By sheer luck, this is possible with the shape I chose for my omni’s.
The omni’s have the shape of a truncated octahedron, consisting of eight hexagons and six squares. The drivers are mounted in four of the hexagons.
Due to this arrangement, there is an empty plane (hexahedron or square) between all the drivers. Damping material can now be placed on these empty surfaces to reduce diffraction.
I took a sheet of melamine foam (4 cm thick) and made the slices from it. This turned out to be quite difficult due to the strange angles and double and triple miter. However, in the end I managed more or less to do so.
Halfway through I realized that only the areas visible from the sweet spot had to be covered. Both speakers were treated in this way.
I didn't bother to take SPL measurements, I was only concerned with the stereo behavior. Because I know of no objective method to measure stereo quality, I just started listening to music.
The result exceeded my expectations. The improvement was immense. The stereo image was sparkling and very detailed. Better than I'd ever heard.
To be honest, I would like to emphasize that I am not an expert in qualitative assessment of loudspeaker systems, nor an experienced stereo listener, but I liked what I heart.
In short, I am now very happy with my omni directional speakers.
My only remaining concern are the WAF aspects.
Jaap
Last edited by a moderator:
How porous is that material? Looking at the basic shape now you have sharper corners which should imply more nasty diffraction... but if that whits stuff is really absorbing/dampening it can work. Is it https://se.rs-online.com/web/p/acoustic-insulation/1034063 ?
//
//
Apparently it works. I don't know how porous this material is. It helps that damping is only really needed where the problems are, namely above approximately 1 kHz.
I bought the material from https://www.akoestiekwinkel.nl/flamex-basic-akoestische-platen
This site also provides some information about absorption. It looks similar to the material on the website you mentioned.
I bought the material from https://www.akoestiekwinkel.nl/flamex-basic-akoestische-platen
This site also provides some information about absorption. It looks similar to the material on the website you mentioned.
Some carefully crafted grilles would greatly help that.
You may also be able to get by with thinner melamine. With thinner materials you get less absorption at lower frequencies, but you may still get the result you want.
That is great work. A few off axis measurements of one speaker 0 deg, 30 deg, 60 deg should show the effect that lead to the improved stereo image. I expect you could create a cylindrical lamp shade using some wire and acoustically transparent cloth and metal screen, add a light bulb above and below the speaker and tell your wife it is a lamp!
I have planned to take measurements. However this requires foam slices on all free surfaces, not just the front. So I will first complete the foam slicing and also take WAF aspects into account as much as possible. Luckily, I have already done some REW measurements on the speakers without foam, so a good comparison should be possible.
I have a feeling that camouflage is not the right way to go. I like functional shapes and think that they are more likely accepted. I am thinking about accentuate the foam slices by, for example, using two types of foam with different shades of gray.
I have a feeling that camouflage is not the right way to go. I like functional shapes and think that they are more likely accepted. I am thinking about accentuate the foam slices by, for example, using two types of foam with different shades of gray.
Interesting suggestion. I will think about it
I took a REW measurement with the foam-covered omni.
A measurement of the omni without foam is also shown for comparison.
Although there was a clear difference in stereo separation, the improvement in SPL is noticeable but not very spectacular.
I suspect that thicker foam will work better but this is not possible without limiting the field of view of the drivers.
Also given the problematic practical implementation, I see foam covering as a dead end for this moment.
A measurement of the omni without foam is also shown for comparison.
Although there was a clear difference in stereo separation, the improvement in SPL is noticeable but not very spectacular.
I suspect that thicker foam will work better but this is not possible without limiting the field of view of the drivers.
Also given the problematic practical implementation, I see foam covering as a dead end for this moment.
A couple thoughts:
1) is your gating/windowing short enough to remove room effects?
2) have you tried measuring the front driver alone, then adding the other drivers one at a time to compare frequency response flatness? I'm not sure what's inherent to the enclosure/baffle shape, what's inherent to the driver, what is because of multiple spaced drivers, etc.
1) is your gating/windowing short enough to remove room effects?
2) have you tried measuring the front driver alone, then adding the other drivers one at a time to compare frequency response flatness? I'm not sure what's inherent to the enclosure/baffle shape, what's inherent to the driver, what is because of multiple spaced drivers, etc.
ad 1) I hope so. Mic distance was 1 meter, window was 2.5 msec
ad 2) I think these kind of measurements are only useful if the results can be
compared with a theoretical model. Unfortunately, the maths of diffraction
effects are way over my head.
However, I did take some measurements at different angles in the horizontal plane.
I'll look it up and post it.
ad 2) I think these kind of measurements are only useful if the results can be
compared with a theoretical model. Unfortunately, the maths of diffraction
effects are way over my head.
However, I did take some measurements at different angles in the horizontal plane.
I'll look it up and post it.
Herewith some measurements I did on the omni's last year. Obviously without the foam slices.
The omn's were rotated in the horizontal plane at 30 degree intervals. Because the omni's are left-right symmetrical, a coverage of 180 degrees was sufficient. The pictures, taken from the microphone position, shows the position of the omni at the respective measurements
Clearly, the combination of diffraction and interference make a mess of the speakers' transfer functions.
Apparently our hearing in general can handle this just fine because it is evolutionarily trained.
However, this practicedness does not apply to stereo hearing; we have known the phenomenon of stereo for less than a hundred years. And there I think lies the problem of omni based on platonic bodies.
The omn's were rotated in the horizontal plane at 30 degree intervals. Because the omni's are left-right symmetrical, a coverage of 180 degrees was sufficient. The pictures, taken from the microphone position, shows the position of the omni at the respective measurements
Clearly, the combination of diffraction and interference make a mess of the speakers' transfer functions.
Apparently our hearing in general can handle this just fine because it is evolutionarily trained.
However, this practicedness does not apply to stereo hearing; we have known the phenomenon of stereo for less than a hundred years. And there I think lies the problem of omni based on platonic bodies.
Only a smooth in-wall mounted driver is diffraction free (in principle) . Do you find that a bliss ? I think not. The space behind the loudspeaker somehow participates in determining the soundstage.
I have also a question about the following:
I have also a question about the following:
Have you tried this yourself ? If so what was the result ?
That's one of the more interesting aspects to me if it's the source of most of the response variations. Obviously at some point you are going to have problems no matter what with spaced drivers, but if you can get them closer together you can probably push the trouble zone much higher in frequency. If you can get that point high enough, the directionality of the sources may help minimize the impact to the measured frequency response also.
Obviously the aesthetics of the speaker may have to change significantly to make that happen, so it may be a non-starter for you just from that standpoint.
Reducing the size of the omni in order to reduce the problems is certainly a possibility.
However, this is not easy. A smaller speaker often means the use of smaller drivers whose bundling will then occur at higher frequencies.
True, the trouble will then shift to a higher frequencies, roughly proportional to the reduction in size , but the problem is not solved.
However, this is not easy. A smaller speaker often means the use of smaller drivers whose bundling will then occur at higher frequencies.
True, the trouble will then shift to a higher frequencies, roughly proportional to the reduction in size , but the problem is not solved.
I was assuming the driver size and angular orientations would remain the same, but they would get closer together; if the majority of the enclosure volume could be shifted away from the drivers themselves.
I've built a few desk speakers where things were really crammed together, so your enclosure looks spacious from here. But it's hard to judge how much room you have between drivers since the angles are different than I'm used to. Maybe things are tighter than they look.
I've built a few desk speakers where things were really crammed together, so your enclosure looks spacious from here. But it's hard to judge how much room you have between drivers since the angles are different than I'm used to. Maybe things are tighter than they look.
You have a problem with indirect drivers, You should change the polarity of this 3 drivers and then measure with both poles. I use ambiental tweeter positioned vertical on normal speaker box, and this tweeter has always!! changed phase, with in phase you have combs in frq plot. Ambiental is crossed by 1-2kHz 6dB and attenuated 6dB for best reproduction.