The small red spheres sold and left the house a few weeks ago. A friend dropped by and had a listen to them. After two minutes he said "My speakers will never sound this good. I'm buying these". So it is time to start on another pair. I have a few 2" Whisper drivers, so I made a print to see how that works. The Prusa slicer makes these tree shaped supports that save filament but look weird. The first print in black looks OK. I think the red looked better. I messed up the driver mount cutout depth a bit as these protrude from the sphere by a few millimeters. I think it will pair well with two 10" Vifa woofers. The black does match the woofer box nicely. These rare earth magnet woofers are great as the woofer and the box combine only weigh 24 lbs.
Very cool approach. Question though how much of the midrange comes from the woofer in the cube? Mid frequencies (vocals esp.) benefit from spheres, and cubes are generally not the best shape for these.
The small red spheres (post #450) with 1" Aurasound Cougar drivers crossed over at 300Hz with 24dB/oct to a single 12" woofer. That actually worked better than I expected. Moving up to twenty of the 2" Aurasound Whisper drivers I hope to get the crossover down to maybe 250 or 200 Hz without sacrificing too much dynamic range. I plan to use three 10" woofers in three stacked boxes pointed in different directions with each sphere. The box I bought is not idea, as I wanted the half cube shape, but these will probably work after I add bracing and stuff them with fiberglass to move all the resonances above 600Hz and damp them. With the woofer box panel and acoustic resonances well above the crossover frequency there will be no audible "box sound" in the midrange. Without bracing or stuffing I knocked on the woofer box wall and this was spectrum measured with a microphone. Two nice panel resonances right about 300Hz and 400Hz. So bracing the panels should move those higher. Unchanged these boxes would sound like crap. A pure 250 Hz sinewave is difficult to locate. If the woofer produces harmonic distortion, the 500, 750Hz harmonics give away the location. So with a low distortion woofer the voices in that range locate to the sphere that is reproducing the recorded harmonics in the voice.
Last edited:
Makes sense, did you make the 24 db / oct. xovers yourself or were able to buy somewhere? I am looking to get hold of 24 db / oct. 100 Hz high pass filters (LR if possible) for some midwoofers.
I use digital crossovers. I find most people have a few amplifiers laying around and a DSP crossover often costs about the same as a quality 4th order passive crossover. AT 100Hz the coils required get large and expensive and tuning a passive crossover requires buying different parts. Have a look at the minidsp.com site. I used two miniDSP 2x4HD boxes and four Dayton APA-150 amplifiers. Each sphere had three channels, front, mid ring and rear, with the forth channel for the woofer. This allowed setting different delays / phase and EQ to produce Omni, Cardioid and forward point source directivity patterns. I recently used the Hypex plate amps with built in digital crossovers. Compared to buying separate amplifiers, crossover box, and cables these are a bargain. Madisound sells them.
Last edited:
I found that there is sort of a upper cutoff frequency for an omni directional speaker dependent on its size. Using the Boxsim speaker design software I modeled different size cubes with six of the same 2" drivers (4 cm cones) centered on each face. As boxsim can't make a sphere this is a pretty good approximation. The radiation pattern repeats every 90 degrees and it's makes for some interesting plots. The first image is the setup for the smallest cube about 6 cm (about 3"), close to the size of the drivers. The second is the setup for the 15 cm cube. I generated larger cubes that were 15 cm ( 6 inch ) and 20" on a side. These are all omni directional at low frequency (uniform color in the plot top to bottom) and then transition based on the size of the cube. The same 2" driver was used in all examples. It is interesting to note that the first deviation in the response is a boost in output at 45 degrees followed by dips in output going up in frequency. The conclusion is that a very small cube or sphere is required to get an even omni-directivity pattern at the highest frequencies. If you don't have a tube amp, use that computer running speaker simulations to stay warm out there.
Above: 3" omni directional cube speaker directivity pattern. A uniform omni-directional pattern up to about 4,000 Hz.
Above: 6" omni directional cube speaker directivity pattern. A uniform omni-directional pattern up to about 1,600 Hz.
Above: 20" omni directional cube speaker directivity pattern. A uniform omni-directional pattern up to 500 Hz.
Above: 3" omni directional cube speaker directivity pattern. A uniform omni-directional pattern up to about 4,000 Hz.
Above: 6" omni directional cube speaker directivity pattern. A uniform omni-directional pattern up to about 1,600 Hz.
Above: 20" omni directional cube speaker directivity pattern. A uniform omni-directional pattern up to 500 Hz.
Last edited:
Diffraction. On cubicle also the edges become sound sources. If it is a sphere there is no edges but sound of every transducer propagates to all directions, all around, meaning that interference starts when half circumference of the structure is about 1/3wl, at 1/2wl there is first interference null and comb filter above up to wavelengths where all the transducers beam, about when wavelength is shorter than diameter of transducer.
If you want omni up to 20kHz the construct must be smaller than 40kHz wavelength, about 8mm, or 1/4". Yeah, construct, including the transducer and any physical structure that comes with it needs to fit on your finger nail.
So, perhaps relax from the requirements a bit, and keep everything as small as feasible. This of course makes small total Sd which means poor low frequency response, which then calls for separate bigger low frequency transucer and structure etc. And it's back to same set of compromises as any loudspeaker system, it is all about sound wavelength, physical size of sound.
If you want omni up to 20kHz the construct must be smaller than 40kHz wavelength, about 8mm, or 1/4". Yeah, construct, including the transducer and any physical structure that comes with it needs to fit on your finger nail.
So, perhaps relax from the requirements a bit, and keep everything as small as feasible. This of course makes small total Sd which means poor low frequency response, which then calls for separate bigger low frequency transucer and structure etc. And it's back to same set of compromises as any loudspeaker system, it is all about sound wavelength, physical size of sound.
The irregular polar response is likely due to the distance between the drivers causing constructive and destructive interference.
I purchased a few 1" outer diameter tweeters last week. Unfortunately the housing is fairly deep. So building that tiny tweeter sphere or cube is a challenge. It would be great to get to 15 kHz with an even pattern using a tiny tweeter moon orbiting the midrange planet.
Because of the symmetry and waves from drivers arriving at the cube edges from both directions it's not clear to me that diffraction is happening as it would with a single driver on a baffle. The software has settings for rounding the baffle edges. I didn't find they had a noticeable effect on this omni simulation, so I think the cube is a good approximation for the sphere in this model.
I purchased a few 1" outer diameter tweeters last week. Unfortunately the housing is fairly deep. So building that tiny tweeter sphere or cube is a challenge. It would be great to get to 15 kHz with an even pattern using a tiny tweeter moon orbiting the midrange planet.
Because of the symmetry and waves from drivers arriving at the cube edges from both directions it's not clear to me that diffraction is happening as it would with a single driver on a baffle. The software has settings for rounding the baffle edges. I didn't find they had a noticeable effect on this omni simulation, so I think the cube is a good approximation for the sphere in this model.
The irregular polar response is likely due to the distance between the drivers causing constructive and destructive interference.
As the centre-to-centre of the drivers rises above a quarter wavelength one starts to get lobing caused by comb filtering. Given the non-panar mounting driver dispersion will play a greater role.
8 sided, vrs 12 sided, vrs 20 sided will get more and more complex in twrms of the lobing.
For reference, combing on a plane (5 driver line array).

dave
I completed a third version of spheres. I was able to cram six two inch drivers into a 4" diameter sphere and the results are very good. The smaller sphere extends the high end dispersion pattern cleanly to about 4 kHz. Above that some equalization works well to smooth the magnitude response variation due to the directivity lobes and produce a good response to above 10 kHz. I demonstrated them at the Pacific NW Audio Society meeting yesterday (www.audiosociety.org) for about 50 people. The speakers were very well received with everyone complementing the sound. I set them up "in the round" with two sets of 20 chairs one in front of the speakers and one behind the speakers, as they are fully symmetrical. With such a large group and large room they didn't benefit from the reflections off side and back walls so the effect was very good, but not as good as it is in my listening room. This time with six of the 2" drivers. I calculated that these will produce an SPL max of 109 dB at 300 Hz limited by the drivers at Xmax. To produce the same 109 dB SPL at 36 Hz requires three 10" woofers with Xmax of 9mm in sealed boxes. I am currently using just one woofer per side and it works pretty well.
I just competed a small two tweeter flying saucer to orbit the small red planet. It looks promising, I have yet to get a crossover put together to test it with the sphere.
I just competed a small two tweeter flying saucer to orbit the small red planet. It looks promising, I have yet to get a crossover put together to test it with the sphere.
These are very interesting and look great too. Can I ask what drivers are these and overall how low bass do you get? I have built two-sphere speakers with added sub for low bass, and four-sphere design with the subs built in each. Would you look to add say a 1" dome tweeter to this above the sphere?
I'm eagerly awaiting the Kirkland Reference version at my local Costco.
Seriously, though, the flying saucers would make excellent surround and rear speakers!
The whole thing would make an awesome system. I wonder, have you compared the spheres to bipolar speakers?
Seriously, though, the flying saucers would make excellent surround and rear speakers!
The whole thing would make an awesome system. I wonder, have you compared the spheres to bipolar speakers?
Six Aurasound Whisper 2" full range drivers. There is no bass from the sphere. A 10" woofer in a separate box handles that. The Whisper has an fs of around 150 Hz. I have the crossover at 300 Hz where the distortion starts to rise. The black UFO has 1" tweeters to cover above 4.5 kHz where the 4" sphere starts to roll off.These are very interesting and look great too. Can I ask what drivers are these and overall how low bass do you get? I have built two-sphere speakers with added sub for low bass, and four-sphere design with the subs built in each. Would you look to add say a 1" dome tweeter to this above the sphere?
I bought a pair of these to use as a center channel speaker...
https://www.madisoundspeakerstore.com/full-range-speaker-kits/markaudio-tozzi-one-kit-pair/
https://www.madisoundspeakerstore.com/full-range-speaker-kits/markaudio-tozzi-one-kit-pair/
I did listen to a pair of low cost Definitive Technologies bipolar speakers a few weeks ago. The rear firing drivers created a less focused but deep , wide and very entertaining presentation sort of like an open baffle speaker with a relatively flat response and very deep bass. The sphere speakers take it a step further sending sound in all directions creating reflections off everything the result being your ears locate the sound as coming from points on a line between the speakers or directly from the speakers sort of floating in space.I'm eagerly awaiting the Kirkland Reference version at my local Costco.
Seriously, though, the flying saucers would make excellent surround and rear speakers!
The whole thing would make an awesome system. I wonder, have you compared the spheres to bipolar speakers?
To get these on the shelf in Costco I would have to build more than five pairs this year. I'm just way too lazy for that. Ha.
There are interesting articles that dispute whether comb filtering really makes any audible difference given how humans sound-to-aural-to-neural processing works...one of them:
https://www.axiomaudio.com/blog/comb-filtering-popular-misconceptions
By Alan Loft:
Alan Lofft was, for 13 years, Editor in Chief of Sound & Vision, Canada's largest and most respected audio/video magazine. He edited Sound & Vision (Canada) until 1996, when he moved from Toronto to New York to become Senior Editor at Audio magazine.
https://www.axiomaudio.com/blog/comb-filtering-popular-misconceptions
By Alan Loft:
Alan Lofft was, for 13 years, Editor in Chief of Sound & Vision, Canada's largest and most respected audio/video magazine. He edited Sound & Vision (Canada) until 1996, when he moved from Toronto to New York to become Senior Editor at Audio magazine.
Hi Alan,
That's some marketing material that they claim supports the design choices for the speaker they build. I see that Axiom Audio makes a large tower speaker they claim to be omni-directional while it obviously cannot be a uniform full range omni-directional source given the widely varying directivity pattern produced by vertically stacked tweeters and midranges, drivers spaced far apart on the cabinet and no drivers on the sides, top or bottom surfaces.
I think it is important to differentiate between between a comb filtering effect measured at one microphone location produced by multiple speaker boxes, each with even directivity and flat frequency response, distributed in a room verses a single speaker with wild variations in frequency response and directivity up and down the scale due to driver spacing and crossover choices for that one box. I think you could imagine that it would be possible to construct the same frequency response at a single microphone with both scenarios, but the listening experience would be very different when listening to a stereo pair.
While trying different crossovers with a friend on his large 3-way tower speakers we discovered an interesting audible effect. With a three way first order crossover, the single speaker produced a small localized sound stage that gave the impression that this single three way speaker was a large source of sound. This was an entertaining almost synthetic localized "stereo" effect from the one speaker due to the large overlap in the response of the three drivers at the two crossover points. Using presets on a digital crossover, were were able to rapidly switch between different completed crossover designs with a remote control. Each crossover produced a measured flat on axis response. With a higher order crossover, such that there was very narrow overlap of the drivers, the sound from one speaker collapsed down to what sounded like a smaller not quite point source. This is good, but not entertaining, hence the invention of stereo recordings. With a high order crossover the vertical lobes in the directivity are much reduced. So in this case the lobes in the directivity and variation in frequency response at different angles were produced by a single multi-driver speaker due to it's first order crossover, but this is a fixed effect with nothing to do with the recording, the room, or speaker placement. So it produces confusing location cues when combine with room acoustics and recorded venue information. I think many manufacturers such as Axiom, Definitive Tech, Wilson Audio, every multi-driver dipole speaker with low order crossovers and others with drivers scattered around the speaker and / or first order crossovers leverage this sort effect to make an entertaining speaker with a large source presence from each single speaker.
All that to say I think that speakers with uneven directivity and frequency response muddy the water for extracting the information from the recording and producing a performance that sounds like it is happening in your room. With the relatively smooth directivity and frequency response I've achieved with the spheres, see post #466, the results have been very unique and the imaging in the room is perceived to be quite precise. Due to
people being very polite it is often difficult to get critical comments from people that listen to my speakers. Often you just get nothing if they are bad.
In the four occasions I have demonstrated the spheres, the first pair sold in five minutes into the listening session, the second pair was praised by the
owner of a recording studio while playing his personal recordings and I was continuously congratulated on the sound by person after person at the meeting with 40 or so attendees last week. So my anecdotal evidence indicates the even response has some merit.
Dana
That's some marketing material that they claim supports the design choices for the speaker they build. I see that Axiom Audio makes a large tower speaker they claim to be omni-directional while it obviously cannot be a uniform full range omni-directional source given the widely varying directivity pattern produced by vertically stacked tweeters and midranges, drivers spaced far apart on the cabinet and no drivers on the sides, top or bottom surfaces.
I think it is important to differentiate between between a comb filtering effect measured at one microphone location produced by multiple speaker boxes, each with even directivity and flat frequency response, distributed in a room verses a single speaker with wild variations in frequency response and directivity up and down the scale due to driver spacing and crossover choices for that one box. I think you could imagine that it would be possible to construct the same frequency response at a single microphone with both scenarios, but the listening experience would be very different when listening to a stereo pair.
While trying different crossovers with a friend on his large 3-way tower speakers we discovered an interesting audible effect. With a three way first order crossover, the single speaker produced a small localized sound stage that gave the impression that this single three way speaker was a large source of sound. This was an entertaining almost synthetic localized "stereo" effect from the one speaker due to the large overlap in the response of the three drivers at the two crossover points. Using presets on a digital crossover, were were able to rapidly switch between different completed crossover designs with a remote control. Each crossover produced a measured flat on axis response. With a higher order crossover, such that there was very narrow overlap of the drivers, the sound from one speaker collapsed down to what sounded like a smaller not quite point source. This is good, but not entertaining, hence the invention of stereo recordings. With a high order crossover the vertical lobes in the directivity are much reduced. So in this case the lobes in the directivity and variation in frequency response at different angles were produced by a single multi-driver speaker due to it's first order crossover, but this is a fixed effect with nothing to do with the recording, the room, or speaker placement. So it produces confusing location cues when combine with room acoustics and recorded venue information. I think many manufacturers such as Axiom, Definitive Tech, Wilson Audio, every multi-driver dipole speaker with low order crossovers and others with drivers scattered around the speaker and / or first order crossovers leverage this sort effect to make an entertaining speaker with a large source presence from each single speaker.
All that to say I think that speakers with uneven directivity and frequency response muddy the water for extracting the information from the recording and producing a performance that sounds like it is happening in your room. With the relatively smooth directivity and frequency response I've achieved with the spheres, see post #466, the results have been very unique and the imaging in the room is perceived to be quite precise. Due to
people being very polite it is often difficult to get critical comments from people that listen to my speakers. Often you just get nothing if they are bad.
In the four occasions I have demonstrated the spheres, the first pair sold in five minutes into the listening session, the second pair was praised by the
owner of a recording studio while playing his personal recordings and I was continuously congratulated on the sound by person after person at the meeting with 40 or so attendees last week. So my anecdotal evidence indicates the even response has some merit.
Dana
Last edited:
I got curious and reviewed Toole, Sound Reproduction Loudspeakers and Rooms, 2nd Edition.Floyd has said that combing looks way worse than it sounds, but it will eat DDR.
dave
Chapter 5, Timbre Changes / Comb filtering , repetition pitch:
Toole: "The effect is audibly obvious if it occurs in an electronic signal path or if there is a single, strong reflection ... "
My interpretation: the delayed sound and original sound come from the same direction is audible.
Toole: "For reflections that arrive from large horizontal angles .... the effect ceases to be a problem"
My interpretation: With two ears listening sounds from different directions that would produce a comb filter
response at a single microphone are not an audible problem for a listener with two ears. I see no reason that
large horizontal or large vertical angles would not be the same.
Chapter 9 Section 9.1.1 Very Audible Differences From Similar Looking Combs drives the point home with an example
describing three cases a), b) and c) where the single speaker case c) producing a comb filtered output "greatly degrading effect".
Toole's example speaks to producing the comb filter effect electronically but the same effect could be produced acoustically with a single
speaker having several drivers at different distances from the observer.
As usual it is possible for people to draw different conclusions from the same source depending on what they are focused on.
I found it too simple of a "test".There are interesting articles that dispute whether comb filtering really makes any audible difference
https://www.intechopen.com/chapters/45612

So much of what is sensitive is defined by our ear shape and head shape.
The 2 mics don't account for this. (search HRTF for more)
- Home
- Loudspeakers
- Multi-Way
- How to build a spherical speaker?