I was inspired by a presentation from Sigfried Linkwitz in which he augured that that omni directional speakers are better suited to our evolutionarily developed hearing system of ears plus brain processing. He argued that it is important that reflections from the surrounding were little hindered because our hearing system also needs this sound information to locate the (virtual) sound source.
To test his idea, I build my own omni directional speakers. Of importance to me was that they were simple to build, had good omni directional properties, did not need extensive testing and were reasonably priced. Hi SPL and dynamic range were of less importance for this test model.
There are several types of (semi) omni directional speakers. In the design described here, four drivers are mounted on the surfaces of a (semi)regular polyhedron. Useful for this project turned out to be the regular tetrahedron and the regular octahedron, whether or not in truncated form (semi-regular). With these shapes, the minimum of four drivers is sufficient.
The shape above is that of a semi-regular tetrahedron (truncated tetrahedron), made up of four regular hexagons plus four equilateral triangles. This shape is the easiest to build. With a rib length "r", the volume is approx. 2.7 x r3. In each of the four hexagons a driver is mounted. The uniformity of the spherical radiation field turns out to be amazingly good ( ± 1.5 dB, up to over 5 kHz in the distant field).
To be continued…
To test his idea, I build my own omni directional speakers. Of importance to me was that they were simple to build, had good omni directional properties, did not need extensive testing and were reasonably priced. Hi SPL and dynamic range were of less importance for this test model.
There are several types of (semi) omni directional speakers. In the design described here, four drivers are mounted on the surfaces of a (semi)regular polyhedron. Useful for this project turned out to be the regular tetrahedron and the regular octahedron, whether or not in truncated form (semi-regular). With these shapes, the minimum of four drivers is sufficient.
The shape above is that of a semi-regular tetrahedron (truncated tetrahedron), made up of four regular hexagons plus four equilateral triangles. This shape is the easiest to build. With a rib length "r", the volume is approx. 2.7 x r3. In each of the four hexagons a driver is mounted. The uniformity of the spherical radiation field turns out to be amazingly good ( ± 1.5 dB, up to over 5 kHz in the distant field).
To be continued…
I am curious to find out what your results will be. On theoretical grounds there should be a problem with sound power. You want the sound power of a loudspeaker to slope downwards. The on axis spl should be straight. An Omni would seem to have a straight sound power curve if flat on axis.
The actual construction:
The first step was to find a suitable driver. Of importance was: - Sufficient frequency range in the high (to avoid tweeters) - Minimum box dimensions (the ideal omni directional sound source is a point source) - A reasonable price (after all, eight are needed per speaker pair)
After some research I chose the DMA70-8, a 2.5 inch full range woofer from Dayton Audio.
For a Qtb of 0.7, this driver requires a volume of about 0.6 liters net in a closed box. The Fb then becomes approx. 160 Hz. The four drivers then require a common volume of approx. 2.7 liters gross. For this, the rib length of the truncated tetrahedron "r" should be 10 cm. The boxes are filled with some damping material.
The material used for the boxes is 4 mm MDF. The hexagons and triangles must be precise to size. In the absence of a saw table, I ordered the material from a laser cutting service.
The baffles are held in place with adhesive tape and then folded together. The resulting v-seams are filled with wood glue. Also on the inside, all seams must be glued to get sufficient strength.
When the boxes are glued together, the drivers are placed, the wiring is fitted and the adapter for the stand is mounted. At a later stage, the connections are incorporated into the adapter.
To be continued….
The first step was to find a suitable driver. Of importance was: - Sufficient frequency range in the high (to avoid tweeters) - Minimum box dimensions (the ideal omni directional sound source is a point source) - A reasonable price (after all, eight are needed per speaker pair)
After some research I chose the DMA70-8, a 2.5 inch full range woofer from Dayton Audio.
For a Qtb of 0.7, this driver requires a volume of about 0.6 liters net in a closed box. The Fb then becomes approx. 160 Hz. The four drivers then require a common volume of approx. 2.7 liters gross. For this, the rib length of the truncated tetrahedron "r" should be 10 cm. The boxes are filled with some damping material.
The material used for the boxes is 4 mm MDF. The hexagons and triangles must be precise to size. In the absence of a saw table, I ordered the material from a laser cutting service.
The baffles are held in place with adhesive tape and then folded together. The resulting v-seams are filled with wood glue. Also on the inside, all seams must be glued to get sufficient strength.
When the boxes are glued together, the drivers are placed, the wiring is fitted and the adapter for the stand is mounted. At a later stage, the connections are incorporated into the adapter.
To be continued….
Completing the construction:
Due to the relatively high crossover frequency of 160 Hz, it is necessary to provide each speaker with its own subwoofer which must be placed under the omni speaker. The subwoofers also serve as a speaker base.
The drivers are as follows connected to the amplifier:
The resistors in the omni’s ensure that the driver which is aimed directly to the listener sounds 3 dB louder than the three other, indirect, drivers. This suggest more space behind the speaker. note: if the speakers are placed far (> 1 m) from the walls, no resistors should be used.
The stands are steel tubes of 12 mm outside diameter. The connecting wires are incorporated into the tube. Omni's, subs and stands are equipped with "tulip" connectors for "quick release" and simple transport.
At the following photo you can see that the direct drivers are aimed at the listener, taking care that only the direct drivers can be seen.
In this setup, only one single reflection from an indirect driver remains. To reduce this remaining single reflection, a piece of damping material has been applied against the wall. Because of this, together with the attenuated indirect drivers, it turned out to be acceptable that the speakers are placed relatively close to the wall behind it.
My personal findings: The speakers have a nice open sound. Seated in the sweet spot, after some time to get used to it, the speakers as well as the listening room disappear from the sound stage (at least with a good stereo recording). What remains is the sound of the musicians in the room in which they play. The stereo image is sparkling and appears 3-dimensional. All this in an ordinary asymmetrical living room that is not equipped with absorbent or diffuse elements. Outside of the sweet spot, which is already quite large, the sound quality is still very good. My feeling is that Linkwitz was right.
Final note:
I think that 4 drivers in a polyhedron is the maximum. This is why: If you, for example, take a dodecahedron with 12 drivers, you will always “see” more than 1 driver from the listening position. The drivers you “see” are all at a slightly different distance from your ear, so you get cam forming due to the different delays (no time alignment). These distortions probably will not be exactly the same for the right and left speaker. And because in particular differences between right and left results in a woolly stereo image, cam forming should be avoided. So, no more than 4 drivers.
Due to the relatively high crossover frequency of 160 Hz, it is necessary to provide each speaker with its own subwoofer which must be placed under the omni speaker. The subwoofers also serve as a speaker base.
The drivers are as follows connected to the amplifier:
The resistors in the omni’s ensure that the driver which is aimed directly to the listener sounds 3 dB louder than the three other, indirect, drivers. This suggest more space behind the speaker. note: if the speakers are placed far (> 1 m) from the walls, no resistors should be used.
The stands are steel tubes of 12 mm outside diameter. The connecting wires are incorporated into the tube. Omni's, subs and stands are equipped with "tulip" connectors for "quick release" and simple transport.
At the following photo you can see that the direct drivers are aimed at the listener, taking care that only the direct drivers can be seen.
In this setup, only one single reflection from an indirect driver remains. To reduce this remaining single reflection, a piece of damping material has been applied against the wall. Because of this, together with the attenuated indirect drivers, it turned out to be acceptable that the speakers are placed relatively close to the wall behind it.
My personal findings: The speakers have a nice open sound. Seated in the sweet spot, after some time to get used to it, the speakers as well as the listening room disappear from the sound stage (at least with a good stereo recording). What remains is the sound of the musicians in the room in which they play. The stereo image is sparkling and appears 3-dimensional. All this in an ordinary asymmetrical living room that is not equipped with absorbent or diffuse elements. Outside of the sweet spot, which is already quite large, the sound quality is still very good. My feeling is that Linkwitz was right.
Final note:
I think that 4 drivers in a polyhedron is the maximum. This is why: If you, for example, take a dodecahedron with 12 drivers, you will always “see” more than 1 driver from the listening position. The drivers you “see” are all at a slightly different distance from your ear, so you get cam forming due to the different delays (no time alignment). These distortions probably will not be exactly the same for the right and left speaker. And because in particular differences between right and left results in a woolly stereo image, cam forming should be avoided. So, no more than 4 drivers.
I'am not sure what you mean by "to slope downwards" . Did the rest of the description answers your question ?
Thank you for the complement.
No, I did not make reference to the Pluto nor any other Linkwitz design. The statement of Linkwitz concerned omni directional speakers in general. By the way, it’s interesting to watch his presentation: “Linkwitz AES Accurate sound reproduction” https://www.google.com/search?q=Linkwitz+AES+"Accurate+sound+reproduction"&sourceid=chrome&ie=UTF-8 , especially parts 2 and 3 are of interest here.
You are right that 4 drivers just pointing into 3-space does not make an omni, but drivers mounted on the surfaces of platonic bodies, in this case a tetrahedron, radiate almost uniform in al directions and make a good approximation of a true omni directional speaker.
No, I did not make reference to the Pluto nor any other Linkwitz design. The statement of Linkwitz concerned omni directional speakers in general. By the way, it’s interesting to watch his presentation: “Linkwitz AES Accurate sound reproduction” https://www.google.com/search?q=Linkwitz+AES+"Accurate+sound+reproduction"&sourceid=chrome&ie=UTF-8 , especially parts 2 and 3 are of interest here.
You are right that 4 drivers just pointing into 3-space does not make an omni, but drivers mounted on the surfaces of platonic bodies, in this case a tetrahedron, radiate almost uniform in al directions and make a good approximation of a true omni directional speaker.
Somewhat similar in terms of drivers firing in differen directions. https://www.diyaudio.com/community/...eal-lets-build-one-that-does-dave-rat.395457/
dave
dave
A handful of us went to that lecture! I pestered him about electrostatics and his Pluto.
https://www.diyaudio.com/community/threads/linkwitz-lecture-in-london-nov-13th.110900/#post-1336884
As you can see a problem occurs at higher frequencies (above 5 k Hz). Due to bundling the spl at the off-axis directions of the drivers decreases, and with that the total sound power. This causes the sound power to slope downwards.
The problem off course is absence of a decent tweeter. A think this needs correction in the next model.
That build looks fantastic. Mounting the speaker pods on rods above woofer boxes looks great. I have been working towards building a full range omni directional speaker and have found that it is very difficult to find a geometry and shallow drivers that can be packed closely enough to be effectively omni directional at high frequency. As Visaton boxsim software can model a cube with a driver on each face I did that. With six 2" drivers mounted on the faces of a 6 cm cube the directivity plot shows a perfect omni-directional pattern up to about 4 kHz where there is a rise in response at 45 degrees off axis of the drivers, So that is pretty good. I hope to find some small high quality tweeters that can be used on the faces of a smaller cube, maybe 2.5 or 3 cm on a side, to extend the omni directional pattern up to maybe 8 or 10 kHz. I tested several of the small Dayton tweeters but found the distortion was a bit high. The search continues.
This axiom (my bold above) - what kind of border requirements does it have to be true? Perhaps it requires a certain directivity to begin with - or this is actually a statement for a speaker directivity - isn't it? You statement using "On theoretical grounds" is about how to build a non omni-directional speaker -not that there is a theoretical problem with omnis - right?
//
Last edited:
Have you looked at these? If you aren't trying to go too low, they are pretty good. Not cheap when you need 6 of them though.
Peerless by Tymphany OX20SC00-04 3/4" Fabric Dome Tweeter
https://www.parts-express.com/Peerless-OX20SC00-04-3-4-Fabric-Dome-Tweeter-264-1002?quantity=1
mattstat: That is a great looking tweeter for sure and the price is actually great. I may have to buy and test one or two.
I'm looking to use the tweeters 4 kHz and up.
I tested the 1" Peerless Tweeter, which doesn't have the rear chamber, and the results were not great. The SPL level on
the graph is not calibrated and the mic was close to the tweeter, 6 inches or so. I was driving maybe half a Watt at the
time so it was playing maybe 85 dB. (I need to calibrate the level some time). It would be great to have < 0.3 % distortion
at normal to slightly strong listening levels.
It had good 2nd and 4th harmonic distortion, but high 3rd harmonic distortion. This might be OK for the rear firing tweeter.
https://www.parts-express.com/Peerless-OC25SC65-04-1-Textile-Dome-Tweeter-264-1018
I need to spend more time testing and less time clicking buy, accumulating untested parts and writing posts. Ha
I bought a pair of SB Acoustics SB14ST-C000-4 Super-Compact Tweeter w/grille - 4 ohm
SB14ST-C000-4 from Madisound and have yet to test them. They look amazing, but have a fairly deep chamber that
will make for a larger cube / sphere than desired and will limit the high frequency.
I'm looking to use the tweeters 4 kHz and up.
I tested the 1" Peerless Tweeter, which doesn't have the rear chamber, and the results were not great. The SPL level on
the graph is not calibrated and the mic was close to the tweeter, 6 inches or so. I was driving maybe half a Watt at the
time so it was playing maybe 85 dB. (I need to calibrate the level some time). It would be great to have < 0.3 % distortion
at normal to slightly strong listening levels.
It had good 2nd and 4th harmonic distortion, but high 3rd harmonic distortion. This might be OK for the rear firing tweeter.
https://www.parts-express.com/Peerless-OC25SC65-04-1-Textile-Dome-Tweeter-264-1018
I need to spend more time testing and less time clicking buy, accumulating untested parts and writing posts. Ha
I bought a pair of SB Acoustics SB14ST-C000-4 Super-Compact Tweeter w/grille - 4 ohm
SB14ST-C000-4 from Madisound and have yet to test them. They look amazing, but have a fairly deep chamber that
will make for a larger cube / sphere than desired and will limit the high frequency.
Last edited:
Listening to my omni directional loudspeakers, and after doing some measurements, I have concluded the following on sound quality:
1. Everywhere in the listening room, the sound reproduction is 'open' and of excellent
quality. This again strongly suggests that for good sound reproduction, all room
reflections must contribute.
2. That the quality of the stereo image, as experienced in the sweet spot, can be improved.
This second conclusion was made from measurements I did on the omni. In particular, the SPL measurement on the listening axis was startling. With the following picture I want to explain what I think is going on.
Keep in mind that the direct driver is pointed to the listener at the sweet spot.
The listening axis to the sweet spot is perpendicular on the plane in which the direct driver ( drawn in green) is mounted. However, in addition to the sound of this direct driver, also sound from the three indirect drivers (drawn in blue) will reach the listener. This is due to diffraction at the corners of the case. However, the sound from these indirect drivers will arrive later and thus interfere with the sound from the direct driver. The result is combing resulting in an undesirable frequency characteristic.
The following picture shows the resulting measured SPL on the listening axis:
The response is not very flat (a clear understatement) and I think is bad for stereo imaging.
What can be done to correct this ?
Looking at the geometry of the octahedron, you can see that the plane in which the three indirect drivers lie is also perpendicular on the listening axis to the sweet spot. However, this plane is about 12 cm behind the acoustic center of the direct driver.
The solution could be to delay the signal of the direct driver by about 12 cm (a few hundred usec).
On forehand it is clear that an active system (with a dsp) is required for the delay.
This will have a lot of construction consequences, but also gives new opportunities to optimize the omni directional speaker system. Therefore, I want to address these issues later.
1. Everywhere in the listening room, the sound reproduction is 'open' and of excellent
quality. This again strongly suggests that for good sound reproduction, all room
reflections must contribute.
2. That the quality of the stereo image, as experienced in the sweet spot, can be improved.
This second conclusion was made from measurements I did on the omni. In particular, the SPL measurement on the listening axis was startling. With the following picture I want to explain what I think is going on.
Keep in mind that the direct driver is pointed to the listener at the sweet spot.
The listening axis to the sweet spot is perpendicular on the plane in which the direct driver ( drawn in green) is mounted. However, in addition to the sound of this direct driver, also sound from the three indirect drivers (drawn in blue) will reach the listener. This is due to diffraction at the corners of the case. However, the sound from these indirect drivers will arrive later and thus interfere with the sound from the direct driver. The result is combing resulting in an undesirable frequency characteristic.
The following picture shows the resulting measured SPL on the listening axis:
The response is not very flat (a clear understatement) and I think is bad for stereo imaging.
What can be done to correct this ?
Looking at the geometry of the octahedron, you can see that the plane in which the three indirect drivers lie is also perpendicular on the listening axis to the sweet spot. However, this plane is about 12 cm behind the acoustic center of the direct driver.
The solution could be to delay the signal of the direct driver by about 12 cm (a few hundred usec).
On forehand it is clear that an active system (with a dsp) is required for the delay.
This will have a lot of construction consequences, but also gives new opportunities to optimize the omni directional speaker system. Therefore, I want to address these issues later.
Yes, I think this is the big insight that eventually will sink in and take the lead. It also jives with normal living rooms which is a nice thing.
This is also my experiance. EQ, digital, is probably the best bet - but all system will have eventually so no problems.
I think you should try to let the direct signal hit the listener 1ms before the diffuse field does - try it!?
Best of luck - you are on the right track as I see it.
//