Speaker is dodecahedron; angle between each driver is 30 degrees; 30 degrees of measurements describes entire speaker. In fact, 15 degrees from a driver axis to corner of dodecahedron describes speaker completely in horizontal plane. Deep notch at 6kHz is completely inaudible with broadband signals.
I referenced from corner point as 0 degrees, and used measurement increment of 2.5 degrees, out to 30 degrees.
Yes, looks pretty bad doesn't it. It is good that they sound a whole lot better than frequency response plots look; and the way they really do look.
I could readily turn them on their sides, and suddenly horizontal response would look good to you, and vertical null would look no worse than poor vertical response of most 2-way, 3-way, or line array speakers. Turning them on their sides would not change the way they sound at all for listener in typical speaker triangle setup.
Did you look at horizontal "off axis" time v frequency sonograms; this was really crux of build exercise from technical standpoint. They demonstrate highly uniform radiation of entire room. Through speaker rotations reflection profiles remain very stable. No forward firing or dipole speaker would produce such results. Couple with actual listening, compared to very well behaved forward firing speakers, be they direct radiator, or waveguide/horn, or dipole system I've demonstrated that much of what people go on about with avoiding this or that reflection is generally nonsense, and that comb filtering as unacceptable is nonsense; at least when confined to shorter wavelengths.
End all be all speaker? Absolutely not. Fantastic prototype, absolutely yes! Would I recommend them to everybody? No way, only those looking to debunk omnidirectional speakers as spacy, for those looking to debunk all early reflections from room as bad, for those looking to debunk ceiling/floor reflections as bad, for those looking to debunk front wall reflections as bad, and most certainly those looking to get a lot of bang for the buck out of small full range drivers.
Last edited:
well ...and it clearly conflicts with the image depth cues that are recorded, and image height cues as well, for that matter, right?
sure, agreed
Why would they?
Besides a "generally guarantee blown out spacy performance" is such a very unspecific, vague statement... meaning nothing more than that You just don't like something (what exactly???), right?
Tsk, tsk.
You take him to "task" on preconceptions at the end, but fall into a preconception yourself at the start.
Tsk, tsk.
You take him to "task" on preconceptions at the end, but fall into a preconception yourself at the start.
where exactly?
that there are (can be) some recorded depth and height cues for phantom sound sources localization is a fact, right?,
that there may be a conflict between those and the depth and height cues from the speakers as physical sound sources is also a fact, right?
because that the floor reflection is a depth and height cue for a sound source localization is a fact too, right?
on the contrary there is no theory explaining any detrimental effect of the ceiling reflection, that's why I asked why.
where are my preconceptions?
Recording methods have no uniformity across much of the field (some standards have evolved, but many are not employed in universally useful way) beyond that most employ microphones. Close microphone techniques are devoid of depth cues providing normal depth cues of a listener as audience; captured cues are from closer than even instrument's player in many cases. To add scenery to such dry recordings ambiance conveying depth is added in variety of manners, ranging from totally synthetic reverberation, or ambiance recorded simultaneously with ambiance microphone(s). Location of ambiance microphone(s) may be flown high, may be boundary microphones, stereo pairs, or really just about anything the recording engineer dreams up. Often now impulse response of desired reverberation space is convolved with direct track to generate reverberation used in the mix. Capture of these impulses is totally up to whim and fancy of recording engineer too.
In stereo recordings mono, reverberation free track as equal signals in both channels appears as phantom source directly between speakers for listener in plane bisecting points of two speakers. At point in plane directly between speakers, source is perceived as centered in listener's head. This hold true for source with any spectrum. For broadband source with flat frequency response, be it white noise or impulsive in nature the perception remains unchanged. Frequency response for this flat source spectrum is not flat as measured at surface of eardrum, as signal is equalized by effects of head, outer ear, ear canal, and the complex impedance of the eardrum. As head is moved in continuous fashion to different points in bisecting plane, perception of change in frequency response is slow and continuous. Changes in perceived frequency response as head is drawn rearwards in direction resulting in speakers appearing as with typical stereo triangle are typically very small. This is not so with real response at eardrum, especially for wavelengths significantly shorter than diameter of head. This is basis that allows identification of a single source speakers location, but with stereo pair and mono source, sound is still perceived as coming from directly between speakers, but now outside of listener's head. Sound does not appear to come from point directly between speakers in the plane of the speakers. Sound is perceived as coming from behind the plane of speakers that is also perpendicular to the plane bisecting the speakers. This is so due to the summing of depth information of each speaker at each ear. For two speaker stereo their is no way around this. The net result is that sound of mono source is roughly same distance behind plane of speakers, as the listener distance is from plane of speakers.
Sure a second set of speakers can be employed to the sides, and behind listener to drawn apparent location of mono source forward, but this is fraught with additional flaws and restrictions on listener location and head orientation.
Additional major flaw of stereo technique is when a mono source is panned all the way to left or right it becomes real source and it's apparent depth and location become that for perception of real speaker's location. So to put a vocalist up front, track needs panning all the way or most of the way to left or right. For vocalist front and center, combinations of EQ, level, and reverberation are added to rest of tracks that increase perception of them as occurring as behind the vocalist. Live with it, use more speakers with added complication, use head phones, or only listen to real live sounds.
All this is before considerations both for reflections not directly associated with listener's body, head, ears and for reflections associated with speaker.
Processing time for high frequencies is shorter than for lower frequencies. Beaming of even small tweeter results in following when pointed at ceiling: Intensity of high frequencies bouncing off ceiling to listener are of sufficient amplitude to become direct sound for listener even though delayed by 3-4ms over direct sound. Intensity of somewhat lower frequencies becomes higher in direct sound, and remain in proportion with strong yet delayed and somewhat lower intensity reflection from ceiling along with the higher frequencies. A new phantom source forms for theses intermediate frequencies that is perceived as below the ceiling and above the speaker. This is phantom image with frequency dependent height; readily describable as smeared or spacey. With horizontal firing tweeter, direct sound is guaranteed to be perceived source direction, and phantom image with second speaker is guaranteed to be in horizontal plane. With wide uniform vertical response, higher frequency reflections from both floor and ceiling tend to have similar delays (unless ceiling is much higher above speaker than speaker is from floor) forming secondary phantom source that is in horizontal plane similar in height with direct sound. At greater listening distance delays of floor and ceiling become shorter. At typical listening distances of 2 meters, delays are short enough and intensities of reflections are high enough that reflections and direct sound are fused together for perceptual processing, and once again, apparent height is determined by real source height.
The ceiling effect is only made worse by our hearing acuity which is sharper near horizontal plane and to the forward direction. Looking up at ceiling in direction of reflection heightens perception of ceiling as true source of sound as perceived intensity of somewhat lower frequencies in window of highest acuity also increase.
With more balanced reflections of floor and ceiling as with forward firing or omnidirectional sources at ear height, looking up or down may increase awareness of floor and ceiling, but confirms to our perception that source is before us.
Omnidirectional speaker is not spacey.
Ceiling firing speaker with restricted radiation behavior is spacey. Sit to close and image is entirely from speaker with phantom image often at or below the floor, sit too far back and sound is spacey room sound, and sit in just the right place and image with frequency dependent height forms with additional dependance on head tilt, making this listening position spacey too.
Why wouldn't our evolutionary development around the ubiquitous existence of ground plane be a feature of our basic hearing perceptual abilities? It certainly isn't auditory stumbling block for perceptions of real sound sources. Why would it be for speakers and phantom sources created between speakers?
In stereo recordings mono, reverberation free track as equal signals in both channels appears as phantom source directly between speakers for listener in plane bisecting points of two speakers. At point in plane directly between speakers, source is perceived as centered in listener's head. This hold true for source with any spectrum. For broadband source with flat frequency response, be it white noise or impulsive in nature the perception remains unchanged. Frequency response for this flat source spectrum is not flat as measured at surface of eardrum, as signal is equalized by effects of head, outer ear, ear canal, and the complex impedance of the eardrum. As head is moved in continuous fashion to different points in bisecting plane, perception of change in frequency response is slow and continuous. Changes in perceived frequency response as head is drawn rearwards in direction resulting in speakers appearing as with typical stereo triangle are typically very small. This is not so with real response at eardrum, especially for wavelengths significantly shorter than diameter of head. This is basis that allows identification of a single source speakers location, but with stereo pair and mono source, sound is still perceived as coming from directly between speakers, but now outside of listener's head. Sound does not appear to come from point directly between speakers in the plane of the speakers. Sound is perceived as coming from behind the plane of speakers that is also perpendicular to the plane bisecting the speakers. This is so due to the summing of depth information of each speaker at each ear. For two speaker stereo their is no way around this. The net result is that sound of mono source is roughly same distance behind plane of speakers, as the listener distance is from plane of speakers.
Sure a second set of speakers can be employed to the sides, and behind listener to drawn apparent location of mono source forward, but this is fraught with additional flaws and restrictions on listener location and head orientation.
Additional major flaw of stereo technique is when a mono source is panned all the way to left or right it becomes real source and it's apparent depth and location become that for perception of real speaker's location. So to put a vocalist up front, track needs panning all the way or most of the way to left or right. For vocalist front and center, combinations of EQ, level, and reverberation are added to rest of tracks that increase perception of them as occurring as behind the vocalist. Live with it, use more speakers with added complication, use head phones, or only listen to real live sounds.
All this is before considerations both for reflections not directly associated with listener's body, head, ears and for reflections associated with speaker.
Processing time for high frequencies is shorter than for lower frequencies. Beaming of even small tweeter results in following when pointed at ceiling: Intensity of high frequencies bouncing off ceiling to listener are of sufficient amplitude to become direct sound for listener even though delayed by 3-4ms over direct sound. Intensity of somewhat lower frequencies becomes higher in direct sound, and remain in proportion with strong yet delayed and somewhat lower intensity reflection from ceiling along with the higher frequencies. A new phantom source forms for theses intermediate frequencies that is perceived as below the ceiling and above the speaker. This is phantom image with frequency dependent height; readily describable as smeared or spacey. With horizontal firing tweeter, direct sound is guaranteed to be perceived source direction, and phantom image with second speaker is guaranteed to be in horizontal plane. With wide uniform vertical response, higher frequency reflections from both floor and ceiling tend to have similar delays (unless ceiling is much higher above speaker than speaker is from floor) forming secondary phantom source that is in horizontal plane similar in height with direct sound. At greater listening distance delays of floor and ceiling become shorter. At typical listening distances of 2 meters, delays are short enough and intensities of reflections are high enough that reflections and direct sound are fused together for perceptual processing, and once again, apparent height is determined by real source height.
The ceiling effect is only made worse by our hearing acuity which is sharper near horizontal plane and to the forward direction. Looking up at ceiling in direction of reflection heightens perception of ceiling as true source of sound as perceived intensity of somewhat lower frequencies in window of highest acuity also increase.
With more balanced reflections of floor and ceiling as with forward firing or omnidirectional sources at ear height, looking up or down may increase awareness of floor and ceiling, but confirms to our perception that source is before us.
Omnidirectional speaker is not spacey.
Ceiling firing speaker with restricted radiation behavior is spacey. Sit to close and image is entirely from speaker with phantom image often at or below the floor, sit too far back and sound is spacey room sound, and sit in just the right place and image with frequency dependent height forms with additional dependance on head tilt, making this listening position spacey too.
Why wouldn't our evolutionary development around the ubiquitous existence of ground plane be a feature of our basic hearing perceptual abilities? It certainly isn't auditory stumbling block for perceptions of real sound sources. Why would it be for speakers and phantom sources created between speakers?
The floor "reflection" doesn't *conflict* with any cues - you don't hear it as a separate source so there is no conflict. (..you only get a conflict with added time. (..an *alteration* effect (perceptual) would be moving the image closer to the reflection which requires a much higher pressure level than the direct sound, it other-wise requires a great deal of time (echo) to get an actual conflict.)
Additionally there is nothing "clearly" about its alteration - it's very much loudspeaker design and room dependent relative to the magnitude of any alteration and the listener's ability to hear that alteration.
The best way to express a vertical reflection (floor or ceiling) under reproduction is just with the freq. response at and very near the listening position: a result oriented perspective.
IF there is a difference from "flat" as a result of the reflected condition (within our constraint), then it creates a "dip" in response. That "dip" and its artifact's effect is dependent on the "Q" of the depression and to an extent below 2-3 db from the average: its "depth". A broader-band "Q" effect at -3db or more tends to shift images of sources backward. Same sort of effect with a digital eq. creating a fairly broad depression in the midrange.
A change in perceptual vertical (image placement) "height" effects are rare - principally because the source (upper mid-to-tweeter) is usually sufficiently far enough away from the floor or ceiling. Further, any effects under this condition that would show up at the listening position are of the extremely narrow "Q" variety (with regular intervals) that result in a typical combing effect - exceedingly similar to other surface reflections (like walls). This affects neither depth nor height localization of images. Note however that this changes IF the reflection is off of a near surface like a table close to the listener - at that point you will likely have freq. artifacts that are more troublesome (but it still doesn't become a competing source).
Last edited:
status-quo-Markus
@tinitus - Markus didn't like the bass of Beolab, problem is that Beolab dealers are not audiophiles, so to speak..., and to make things worse they are told to pump up the bass correction at max position as it pleases generally the consuming crowd, and the bass was probably screwing up the rest
I've tried a bunch of different directivity options. I have a set of Summas with 90 degrees of coverage, some Vandersteens that are close to 180 degrees due to their very small baffles.
I also have some omnipoles.
To my ears, the cardioid-like radiation of the Beolabs sounds the most natural. Basically these speakers lie on a spectrum. On one end of the spectrum is the omnipoles, with very diffuse radiation. I find that omnipoles make the soundstage too diffuse. Very directional speakers, on the other end of the spectrum, can be ruthlessly revealing, but the soundstage can suffer.
Cardioid-like speakers like the Beolab seem to occupy a 'sweet spot' in the middle. Not too diffuse, not too small. Not too revealing, not too forgiving.
The downside for me was dynamics; due to the wider radiation angle the speaker couldn't get as loud as effortlessly as my reference.
The Acoustic Lens isn't really cardioid at all (though I guess "like" is a good modifier):
Moulton Laboratories :: A new loudspeaker design
Most "omni's" would like to have such a wide horizontal dispersion pattern. I'd say it's a solid "radial" design for a 160 degree horizontal "window".
The upper tweeter lens does have significant reduction in output for 40 degrees centered on 180 above 5 kHz. Not exactly surprising considering that at 5 kHz the wave's length is little less than 3 inches and the lens's width is about that as well.
It doesn't provide any localization cue for reproduced images. It may however alter localization cues reproduced by the loudspeaker.
As for the loudspeaker being a sound-source, floor reflections *could* become a cue. (..but they could also just as easily be something that reduces localization of the loudspeaker.)
Can You please cite any scientific references for this theory?
You mean by the tweeter position? Ypu think it's good? Would You call it fidelity or realism etc.?
Clearly You describe Your subjective impressions of a certaint setup. Am I right?
Still I don't know what "spacey" means.
Neither I know what is the objective argumentative force of Your subjective impression You got with a certain setup for the rest of us here.
Because this existence is ubiquitous in the environment in which we evolved? And exactly therefore the ground reflection is a spatial cue that our hearing relies on.
Isn't it a reasonable conlusion?
That's the problem in fact.
Because phantom sources are created by means of two real sound sources? And perception of those real sound sources interferes with the perception of the phantom source? And therefore sometimes we just hear the two speakers instead of the phantom source (the sound is "tied to the speakers", no "disapperaing act")?
Last edited:
It shouldn't be (in either instance), or at least not a significant one that can't be dealt with.
Last edited:
Check out my measurements at my thread named "28 Days Later."
Beolab doesn't advertise the fact that it's cardioid, but it definitely is.
This was a real eye-opener for me; until I measure the lens I'd assumed that it's patter was closer to omni.
An externally hosted image should be here but it was not working when we last tested it.
The secret sauce is the baffle. The lens gets all the attention, but the lens is only working over a narrow bandwidth. About one octave. The baffle behind the driver, and the baffles above and below the driver, are playing a much larger roles in the radiation. Basically the 'cymbal' shaped baffles constrain the loudspeaker radiation into a narrow vertical beam, and the baffle behind the radiator constrains it into 180 degrees. Due to the elliptical shape, as pattern control is lost, the transition is very gentle. (Very little diffraction, and a slow transition from 180 to 360.)
Here's an example of how well the lens works.
This is my clone of the Beolab lens, using a compression driver.
Measurements are 0, 45, and 90 degrees off axis. Grey curve is 180 degrees off axis - behind the speaker. See how the response curve mirrors the "on axis" response, but is strongly attenuated? Very similar to a cardioid.
It's a neat trick, and I've never seen a loudspeaker where the off-axis radiation mimicked the shape of the on-axis so closely. Even speakers with waveguides tend to suffer from high frequency attenuation. The Beolab lens isn't quite as smooth as a waveguide, but the response shape is incredibly consistent. I think this contributes to it's natural sound "in room."
Last edited:
Yea, I consider the "baffle" to be part of the "lens". (..it was what I was referring to with wavelength at 5 kHz up.)
The loss in output at 180 degrees in your example isn't anywhere close to the loss in output of a cardiod. It does have some similarity however, but the magnitude just isn't there. But "like" covers that.
Last edited:
I'd love to see more data on cardioids.
Seems like dipoles and waveguides are currently the darlings of the DIY set, but cardioids can sound very very good in my experience.
I've stumbled across a handful of projects online, but they're few and far between.
Post 8:
http://www.diyaudio.com/forums/multi-way/142691-adventures-cardioid.html#post1808323
Is about as close to a cardiod as you'll get in a fairly standard speaker. Note that it's only measured out to 90 degrees, but loss starts occurring around 15 degrees. The Moulton polar spectorgram is showing almost no loss out to 60 degrees, and not much further out to 80 degrees.
I don't think it's really viable in this format.. unless you going with line-array or you don't care about power/excursion/distortion.
Last edited:
This is actually about microphones but basics are same
Microphone Polar (Directional) Response
Cardioid pattern is used eg. in some Gradient and Amphion speakers and Kimmosto has done some studies and projects with them. DIY archive of Kimmo Saunisto It is very good for bass and mid because of minimal backwards radiation, but it is quite difficult to control those long waves!
Microphone Polar (Directional) Response
An externally hosted image should be here but it was not working when we last tested it.
Cardioid pattern is used eg. in some Gradient and Amphion speakers and Kimmosto has done some studies and projects with them. DIY archive of Kimmo Saunisto It is very good for bass and mid because of minimal backwards radiation, but it is quite difficult to control those long waves!
Then when could it be problem? What's the nature of the problem? How would You deal with it?
I have auditioned the Beolab 5 recently in two different setups and was quite disappointed. Especially for 18 grand a pair, ridiculous.
The phantom image was quite weak up to non-existing.
I think they were not set up correctly in the B&O shop. Both pairs were pushed right up to the front wall and there they really $ucked. I asked the B&O guy if that were an issue and he said no because of the radiation pattern. I also verified that it is 180° radiation. But he did not want to re-arrange them and only demo'ed the 2nd setup with the same issues.
Tonality was average.
P.S. Personally, I would aim for a hyper or super cardioid. Unfortunately, this is not simply done on a few weekends. The mid range seems most challenging.
The phantom image was quite weak up to non-existing.
I think they were not set up correctly in the B&O shop. Both pairs were pushed right up to the front wall and there they really $ucked. I asked the B&O guy if that were an issue and he said no because of the radiation pattern. I also verified that it is 180° radiation. But he did not want to re-arrange them and only demo'ed the 2nd setup with the same issues.
Tonality was average.
P.S. Personally, I would aim for a hyper or super cardioid. Unfortunately, this is not simply done on a few weekends. The mid range seems most challenging.
Last edited: