KEF/Coaxial 2 Ways vs 3 way point source qualities

[Torin Krell]

Skimming over this thread I don't think that there is any mention so far that the best coaxial speaker drivers, with appropriate crossovers, are also coincident. Coaxial and coincident drivers, with appropriate crossovers, would have their drivers aligned in space and time over a broad range of angles minimizing / eliminating phase issues and maximizing inter-driver coherence. Many or most implemented coaxial drivers are actually not fully coincident although many come close. This is likely obvious when the tweeter horn protrudes significantly above the woofer but can also continue to be an issue even when the drivers seem to be aligned appropriately.

 

[thewas]

We have to separate 2 things here: The theory of Blauert regarding boosted frequency bands defining a perceived angle and Zwickert/Fastl´s loudness curve, are solid scientific findings confirmed over and over again by binaural measurements. A variety of HRTF-based algorithms such as virtual immersive surround via earphones or lots of virtual surround algorithms are based on these findings. They would not work, if the underlying groundwork would be wrong and not applicable to reverb in a listening room.

The other thing is the conclusion from that drawn by MEG that the directivity index of a speaker should not be constant but engineered to serve the aforementioned scientific findings. It would basically mean that frequency bands drawing the localization of reverb away from the stereo base, particularly the 0.8-2K band, should be underrepresented in the diffuse field, while the 2.5-5K should be overrepresented, i.e. lower directivity index over an octave.

Yes, the first one is a solid theory but its application on loudspeaker tuning is just a thesis from a company for its own marketing. By the way their most famous monitor 901K shows wide directivity both on those bands and only a narrowing between 2-2.5 kHz so is not consequent to those statements.

In numerous listening tests I found the theories of Earl Geddes, Siegfried Linkwitz, Andrew Jones and some others widely confirmed, that a constant directivity and even indirect field frequency response in the critical bands is best.

Can you post concrete quote of their theories? In the end our personal bias is just an anecdotal statement without generalisable weight.

This does not mean the reverb field should be colorated the opposite way with underrepresented brillance/treble in the reverb, like suggested by supporters of the Harman curve, or presumably some people at KEF. In numerous listening tests I found the theories of Earl Geddes, Siegfried Linkwitz, Andrew Jones and some others widely confirmed, that a constant directivity and even indirect field frequency response in the critical bands is best.

Note that combining neutral direct sound with neutral indirect sound is not only a theoretical, intuitive ideal. It is also the easiest way to have home listening conditions which are closer in terms of tonal balance to the average studio conditions, where lower level of reflections, better room treatment and more frequency-independent absorption is standard.

Neutral indirect sound is not happening on most loudspeakers and typical home setups, usually at higher angles or distances the sound field gets darker.

I never found any reproduction setup with dominant midrange/presence reverb and attenuated diffuse field above 2.5K anyhow convincing. It always leads to dull reverb and unnatural ambience as I have found it in numerous listening tests. So could you give me any solid reason why this should be preferrable?

Why should I give you a reasoning for your own subjective personal findings? In contrary in real documented blind listening tests from Harman typical home loudspeakers with smooth and increasing directivity scored highest.

Loudspeakers which have been designed specifically to meet the ideal of constant directivity, might be more expensive and less compact, that is true.

Nevertheless there are examples of affordable speakers to come much closer to the ideal of constant directivity at least between 800 and 8,000Hz where it is most required (and least likely to be correctable by EQ). I have heard examples by Wharfedale, Focal, Technics, Elac, Paradigm, Dali, to name some widely available brands.

In the end our personal experiences are an anecdotal statements without generalisable weight.

 

[aarons915]

Why is it, that you speak for others? There is no 'our auditory system'! Mine is different from your's simply because I'm connected to my senses in a very personal way. I'm not 'following the science' of treating humans as machines.


Why is it that you ever once again drag the playback into fear of missing out, uncertainty about technologies, and doubt in the truth of it?

As I said already, to compare KEF's R3 and LS50 is nothing that would reveal any other than the R3 to be vastly superior in every aspect. Now the claim is introduced that the R3 is less coaxial, but you refuse to acknowledge that coaxial is a rarity anyway. You do not supplement your bold claim by any technological reasoning than observing, visually, an asymmetry in the design of the R3 - nothing, not the least of any engineering insight. A 'might be', and 'I hear that' in sighted, anecdotal auditions as a passer-by. It is not about 'psychacoustics', definitely.

This got remastered, on Tidal obviously: The Birthday Party, She's hit (org/ 1982) I loved it during my last visit to the U.S. What a drag, the psyco Cold Turkey grime is all gone ... think that matters a lot more today.

What lol? Humans are humans, listening tests have shown that regardless of age, gender, race, etc, we mostly prefer the same sound. You should be able to differentiate where a sound comes from, whether that's to the left, right or behind you, correct?

The R3 is not superior in every aspect, there are plenty of people who prefer the LS50 with subs to the R3 with subs, speakers are basically a set of tradeoffs, you don't just automatically gain in every category simply because you move up in price.

 

[Arindal]

the first one is a solid theory but its application on loudspeaker tuning is just a thesis from a company for its own marketing.

No. It has nothing to do with marketing or one specific company. The correlation between tonality and perceived angle on the sagittal plane is widely confirmed by any research and development that has been done in the field of HRTF and virtual binaural algorithms. Delivering a reverb field in the listening room roughly matching the tonality of the reverb on the recording, is not even a matter of ´loudspeaker tuning´, but just common sense applied on reproduction.

As mentioned, I do not agree with MEG that loudspeakers should have built-in correction for diffuse/direct field imbalances. Neither should they have the opposite as an unwanted side-effect of increasing directivity.

Can you post concrete quote of their theories?

I found this approach by the late Siegfried Linkwitz to be a good introduction:

Link to Linkwitzlabs

Room Acoustics and directivity

There is a lot more on this page to explore, some of his AES papers are also useful. What Linkwitz, Griesinger and Geddes were laying out, is not ´my personal bias´.

Neutral indirect sound is not happening on most loudspeakers and typical home setups, usually at higher angles or distances the sound field gets darker.

Yes, the dull tonality of indirect sound and reverb perceived as detached from the phantom sources, is the typical flaw i am talking about and which I personally find unacceptable.

That a more balanced reverb is not happening with all speakers on the market, does not mean that it is not a reasonable goal or indication of solid reproduction. I am not saying the indirect sound should be of linear response over the whole audible frequency range, surely some compromises are acceptable, and the goal is not chasing the flattest horizontal response graph.

As mentioned, the frequency bands 0.8-8K are in my understanding the most important as our auditory system is most sensitive to tonal balance and localization here. The neighboring octave 400-800Hz also plays a role, particularly for reverb tonality, so it should not be as far off and a doublecheck is reasonable if imbalances in this band can be countered with EQ.

Loudspeakers following this concept, are widely available in the meantime, and from the companies I have listed, even pretty affordable ones.

In contrary in real documented blind listening tests from Harman typical home loudspeakers with smooth and increasing directivity scored highest.

To apply your initial statement would mean drawing the conclusion that these subjective scores originate from one single company only and therefore can be filed under marketing ;-)

I would not go so far, but nevertheless ´scored highest´ means nothing else than subjective assessment of preference linked to the specific recordings and acoustics of the test setup. I asked Dr. Toole about these tests, and his answer was that Harman research has never tested the impact of directivity in isolated form, nor tested any speaker representing the idea of constant directivity. They also never asked specifically for imaging-related verdicts. According to him, the last such comparison was published in 2008, and the point was always to test the subjective overall preference for existing loudspeaker models on the market, not theories on loudspeaker design. So the question was ´do you subjectively like the sound of loudspeaker A, B, C or D the most with presented recordings?´.

That does not say anything about which directivity concept comes closer to a theoretical ideal and to average mixing/mastering conditions alike. The practical development of constant directivity speakers just started after the last Harman preference test had been published. If I am not mistaken, the whole research facility is not existing anymore, so maybe it is time to get over it.

 

[Arindal]

The R3 is not superior in every aspect, there are plenty of people who prefer the LS50 with subs to the R3 with subs

That does not come to me as a surprise, but the question is for which aspect of sound quality would one prefer the LS50?

 

[thewas]

No. It has nothing to do with marketing or one specific company. The correlation between tonality and perceived angle on the sagittal plane is widely confirmed by any research and development that has been done in the field of HRTF and virtual binaural algorithms. Delivering a reverb field in the listening room roughly matching the tonality of the reverb on the recording, is not even a matter of ´loudspeaker tuning´, but just common sense applied on reproduction.

The reverb of a recording can be very varying, or do you mean the reverb at the studio monitoring?

As mentioned, I do not agree with MEG that loudspeakers should have built-in correction for diffuse/direct field imbalances. Neither should they have the opposite as an unwanted side-effect of increasing directivity.

Exactly.

I found this approach by the late Siegfried Linkwitz to be a good introduction:

Link to Linkwitzlabs

Room Acoustics and directivity

There is a lot more on this page to explore, some of his AES papers are also useful. What Linkwitz, Griesinger and Geddes were laying out, is not ´my personal bias´.

Where exactly do they claim what you say?

As mentioned, the frequency bands 0.8-8K are in my understanding the most important as our auditory system is most sensitive to tonal balance and localization here. The neighboring octave 400-800Hz also plays a role, particularly for reverb tonality, so it should not be as far off and a doublecheck is reasonable if imbalances in this band can be countered with EQ.

Loudspeakers following this concept, are widely available in the meantime, and from the companies I have listed, even pretty affordable ones.

First of all the thesis of the 0.4-0.8-8k band is your own one, I haven't seen anything in literature about it, second I think you just see what you want to see as those brands you mentioned above vary a lot and often even between their models and many of those behave less constant in that band compared to ones you think do poorly there.

To apply your initial statement would mean drawing the conclusion that these subjective scores originate from one single company only and therefore can be filed under marketing ;-)

You really don't want to put the research Toole and the claims of Kiesler on a similar level, do you?

I would not go so far, but nevertheless ´scored highest´ means nothing else than subjective assessment of preference linked to the specific recordings and acoustics of the test setup. I asked Dr. Toole about these tests, and his answer was that Harman research has never tested the impact of directivity in isolated form, nor tested any speaker representing the idea of constant directivity. They also never asked specifically for imaging-related verdicts. According to him, the last such comparison was published in 2008, and the point was always to test the subjective overall preference for existing loudspeaker models on the market, not theories on loudspeaker design. So the question was ´do you subjectively like the sound of loudspeaker A, B, C or D the most with presented recordings?´.

That does not say anything about which directivity concept comes closer to a theoretical ideal and to average mixing/mastering conditions alike. The practical development of constant directivity speakers just started after the last Harman preference test had been published. If I am not mistaken, the whole research facility is not existing anymore, so maybe it is time to get over it.

It is the best we have till better one is done and our current anecdotal stories and personal experiences are not enough to just get over it. By the way the concept of constant directivity is very old and in documented blind listening more people preferred a Revel with typical home hifi loudspeaker increasing directivity in the band you mention to a rather constant one JBL M2.

 

[Heinrich]

What lol? Humans are humans, listening tests have shown that ... there are plenty of people who prefer the LS50 with subs to the R3 with subs, speakers are basically a set of tradeoffs, ...

What to say? You're not the only one to communicate a very individual, yet common take on audio. And that is, it is a tradeoff anyway, and so you're obliged to not trust it, longing for the better in an - by definition - endless iterative process.

The piece of music I mentioned wasn't just thrown in. As soon as the playback system reveals the missing art in the "remastered" version, I think, you're there, end of journey. Ironically the art was in the imperfections according to audio metrics. A proof by the negative. Question is, if you're there at the end of the journey, where to go? Drop the hobby of showing the notorious expert's dissatisfaction?

Regardless, your FUD (fear/uncertainty/doubt) in regard to the R3 and siblings is not grounded. It's just armchair speculation drived from, deliberately, not taking basic engineering into account. At least, you didn't communicate any (!) reasoning that would refer to technical design parameters (show a basic simualtion, show measurements, develop a scheme that allows others to replicate your findings, etc).

And again @Arindal and @thewas debate the least details in directivity, deliberately ignoring the room acoustics, covering all this mess up.
And I stop complaining, no bad feelings!

 

[thewas]

At least, you didn't communicate any (!) reasoning that would refer to technical design parameters (show a basic simualtion, show measurements, develop a scheme that allows others to replicate your findings, etc).

I had shown the less continuous directivity of the R3 in #18.

And again @Arindal and @thewas debate the least details in directivity, deliberately ignoring the room acoustics, covering all this mess up.

And again these are two different topics, poor acoustics are not an excuse for poor loudspeakers and people recognize and prefer good loudspeakers (thus with smooth on and off-axis sound) even in normal non-acoustically treated living rooms.

 

[Heinrich]

And again these are two different topics, poor acoustics are not an excuse for poor loudspeakers and people recognize and prefer good loudspeakers even in normal living rooms.

And again obsessively referencing other peoples' preferences. Your argument doesn't hold anyway - apparently there's a stop sign installed? Directivity and room acoustics interact. That connection is ignored in issueing blanket statements on poor/good/normal.

Not to mention that it distracts from the actual topic, which could have been clarified "scientifically" to the best of all readers, even the original poster. Is there a prob/ with a bass/mid supplementing a coax at an x/over around 350Hz? How to get into a reasonable evaluation? The how to! It is a bit into educating - people, but ... is it that bad?

 

[thewas]

And again obsessively referencing other peoples' preferences.

Where did I talk about such?

Your argument doesn't hold anyway - apparently there's a stop sign installed? Directivity and room acoustics interact. That connection is ignored in issueing blanket statements on poor/good/normal.

No, your argument doesn't hold and is off topic, we are comparing here two loudspeakers, not two rooms.

Not to mention that it distracts from the actual topic, which could have been clarified "scientifically" to the best of all readers, even the original poster.

I showed anechoic measurements which support the difference/preference, where is your concrete "scientific" contribution in this topic? You just keep niggling without contributing anything substantial.

 

[Heinrich]

And again obsessively referencing other peoples' preferences.

Where did I talk about such?

There you are, just the post before:

... not an excuse for poor loudspeakers and people recognize and prefer good ...

See how it goes down the drain? So many blanket statements.

Look, I give every every excuse for poor speakers, but no excuse for bad hearing aids. The often called out 'psychacoustics' is made for the latter, not for stereo.

I'm a realist, logically and naturally an objectivist, a materialist even- I have to eat. Exactly because of that I dismiss your approach. It is Idealism, that makes up things.

Audio is decidedly not virtual reality, no 'as if', it involves imagination instead. The latter is subjective, and, at least for me not sparked in auditions, fueled by a few seconds of 'Fast Car'. It was a bold and correct move by Toole to shift to 'preference', but for very obvious reasons it isn't quite there.

Please, anyway, won't you refrain from at nauseam debating that directivity stuff everwhere? I think I gave a hint on how to evaluate the original topic theoretically and practically before, quite in the beginning. And now I see it all drowned in - as I feel it, utter nonsense.

 

[thewas]

See how it goes down the drain? So many blanket statements.

No, these are the results of the research of Toole and other, I don't think anyone who is seriously involved with the the field of sound reproduction denies the importance of loudspeaker directivity.

Please, anyway, won't you refrain from at nauseam debating that directivity stuff everwhere? I think I gave a hint on how to evaluate the original topic theoretically and practically before, quite in the beginning. And now I see it all drowned in - as I feel it, utter nonsense.

You are not the prophet giving peasants hints, if you want to seriously contribute to a discussion offer concrete theses with sources, the rest is rather nonscene.

 

[aarons915]

That does not come to me as a surprise, but the question is for which aspect of sound quality would one prefer the LS50?

It's literally the point of this thread and the original post, the R3 do almost everything better than the LS50 except for creating a point source since the sound literally emanates from 2 points. So you have 2 options:

3 way KEF crossed over at 80Hz no problem but not a perfect point source since they have a roughly 500Hz crossover from the UniQ to woofer

2 way KEF crossed in the 120-150Hz range, still able to play loud but a more pure point source. The main point is you can do nothing about the blurring of the midrange between the UniQ and the woofers in the 3 ways but in the 2 ways the only limitation is distortion in the bass frequencies so you can cross them over higher and get the best of both worlds.

 

[Arindal]

the R3 do almost everything better than the LS50 except for creating a point source since the sound literally emanates from 2 points.

While it is correct that you have 2 points of sound emanation with the R3, chances are pretty high that the woofer under anechoic conditions can not be precisely localized as a separated source, nor would it disturb the impression of a stable localization plane for stereo. The horizontal difference in angle is 0 (for which the human ear is sensitive), the vertical one does not exceed 5 degrees (for which it is insensitive) and the frequencies which solely the woofer is reproducing, are below 400Hz.

The main point is you can do nothing about the blurring of the midrange between the UniQ and the woofers in the 3 ways but in the 2 ways the only limitation is distortion in the bass frequencies

I again recommend to do a localization test under nearfield or anechoic conditions to verify that by excluding external factors like reflections and reverb. If the latter are responsible for different imaging which you label ´blurring´, directivity and tonality under certain window angles causing reflections, are a likely cause, and that is where 2-way and 3-way are really showing differences.

And yes, you can do something about it, treating the room and reducing the listening distance.

A 2-way plus subwoofer would solve the problem but most likely would not eliminate audible differences caused by directivity. This is a key point - sorry @Heinrich for digger deeper into this, but it is practically relevant in case one compares 2-way to 3-way.

Exactly.

So, if we agree that increasing directivity index boosting particularly the bands below 2K in the reverb and underrepresenting higher ones (like exemplarily seen with the LS50 Meta), is an inherent manipulation of the reverb field and should be avoided, why should it be an ideal and, as you claim, proven by research to be superior? If MEG´s directivity manipulation is not desired, KEF´s 2-way one should be even more undesirable as it spans over broader bands and is more pronounced in d.i. level.

First of all the thesis of the 0.4-0.8-8k band is your own one,

No. The 400 Hz threshold is the lowest frequency from which on we can perceive vertical localization solely by HRTF in interaurally identical signals (that is the reason why I doubt the woofer is localizable as a separated source). 800 Hz are in scientific literature described as both the transition between interaural time differences being perceived as phase differences lower than 180 to separated periods with higher frequencies (which has lots of implications on perceived imaging based on interaural time delays) as well as the transition between typical HRTF diffraction FR signaling frontal source localization and rear one.

You really don't want to put the research Toole and the claims of Kiesler on a similar level, do you?

No, i don´t. One experiment was repeatedly executed with acoustic music recordings and people who knew about recording them, testing only one aspect of sound reproduction (directivity) and keeping all others identical. While the other one is just a random poll of which commercially available product among outdated, differently flawed products, people like more/find less annoying in recordings, a room and test setup unknown/unfamiliar to them. While from the test design it is already obvious which one is a useful scientific finding, I have also reproduced similar tests with participants and myself, and found the obvious conclusion fully verified, that significantly colorated reverb is perceived as, well, significantly colorated.

Let´s be clear here: a very compact 2-way coaxial design bears compromises. From the design, particularly crossover frequencies and woofer arrangement, KEF has obviously done something to counter the aforementioned problem of lower-midrange-heavy reverb in their 3-way (particularly floorstander) models, even if they might not publicly agree to the whole design philosophy.

I am not mentioning directivity as an off-topic thing, but a core difference between the 2- and 3-way models and why there is no reason to believe that the 2-way are better regarding imaging.

 

[thewas]

While it is correct that you have 2 points of sound emanation with the R3, chances are pretty high that the woofer under anechoic conditions can not be precisely localized as a separated source, nor would it disturb the impression of a stable localization plane for stereo. The horizontal difference in angle is 0 (for which the human ear is sensitive), the vertical one does not exceed 5 degrees (for which it is insensitive) and the frequencies which solely the woofer is reproducing, are below 400Hz.

....

I again recommend to do a localization test under nearfield or anechoic conditions to verify that by excluding external factors like reflections and reverb. If the latter are responsible for different imaging which you label ´blurring´, directivity and tonality under certain window angles causing reflections, are a likely cause, and that is where 2-way and 3-way are really showing differences.

But exactly the avoidance of lobing under some angles is the second advantage of real coincidential source and what matters in listening under real listening rooms, so testing such under rather anechoic conditions isn't expedient for such.

So, if we agree that increasing directivity index boosting particularly the bands below 2K in the reverb and underrepresenting higher ones (like exemplarily seen with the LS50 Meta), is an inherent manipulation of the reverb field and should be avoided, why should it be an ideal and, as you claim, proven by research to be superior? If MEG´s directivity manipulation is not desired, KEF´s 2-way one should be even more undesirable as it spans over broader bands and is more pronounced in d.i. level.

No, the LS50 Meta as I have shown has smoothly increasing directivity (index) over the whole audio frequency and thus cannot be compared to some local deviations and kinks.

No. The 400 Hz threshold is the lowest frequency from which on we can perceive vertical localization solely by HRTF in interaurally identical signals (that is the reason why I doubt the woofer is localizable as a separated source). 800 Hz are in scientific literature described as both the transition between interaural time differences being perceived as phase differences lower than 180 to separated periods with higher frequencies (which has lots of implications on perceived imaging based on interaural time delays) as well as the transition between typical HRTF diffraction FR signaling frontal source localization and rear one.

Again, your extrapolation from those about that the DI in those should be flat is nothing found in literature.

No, i don´t. One experiment was repeatedly executed with acoustic music recordings and people who knew about recording them, testing only one aspect of sound reproduction (directivity) and keeping all others identical. While the other one is just a random poll of which commercially available product among outdated, differently flawed products, people like more/find less annoying in recordings, a room and test setup unknown/unfamiliar to them. While from the test design it is already obvious which one is a useful scientific finding, I have also reproduced similar tests with participants and myself, and found the obvious conclusion fully verified, that significantly colorated reverb is perceived as, well, significantly colorated.

No, this experiment with no documentation is at this point just a claim of an anonymous person in the internet vs. documented and even in AES reviewed papers written ones.

Let´s be clear here: a very compact 2-way coaxial design bears compromises. From the design, particularly crossover frequencies and woofer arrangement, KEF has obviously done something to counter the aforementioned problem of lower-midrange-heavy reverb in their 3-way (particularly floorstander) models, even if they might not publicly agree to the whole design philosophy.

The directivity of the low frequency range of a single coindicential driver is just dictated by the baffle size. But the LS50 isn't a big compromise in directivity form and smootheness as its closer to their flagship Blade 2 Meta then the R models.

I am not mentioning directivity as an off-topic thing, but a core difference between the 2- and 3-way models and why there is no reason to believe that the 2-way are better regarding imaging.

In the end its probably a mix of directivity, single point source and less edge diffraction.

 

[aarons915]

While it is correct that you have 2 points of sound emanation with the R3, chances are pretty high that the woofer under anechoic conditions can not be precisely localized as a separated source, nor would it disturb the impression of a stable localization plane for stereo. The horizontal difference in angle is 0 (for which the human ear is sensitive), the vertical one does not exceed 5 degrees (for which it is insensitive) and the frequencies which solely the woofer is reproducing, are below 400Hz.

I again recommend to do a localization test under nearfield or anechoic conditions to verify that by excluding external factors like reflections and reverb. If the latter are responsible for different imaging which you label ´blurring´, directivity and tonality under certain window angles causing reflections, are a likely cause, and that is where 2-way and 3-way are really showing differences.

And yes, you can do something about it, treating the room and reducing the listening distance.

I know the theory shows that we shouldn't be able to locate frequencies under 500Hz, especially being so close to the midrange driver but I'm telling you it's pretty easy to spot once you A/B them, by the way I do that in mono with a single speaker of each so there are fewer reflections and it's easier to spot differences. Years back when I had the R3 for awhile I bought a pair of Neumann KH120 to compare and noticed the same thing, even though the Neumann are not coaxial, they sounded more like a point source than the R3 in my comparison. My theory at the time was since all of the fundamental frequencies and especially the vocals were coming from the midwoofer, it sounded like a point source. I ended up not liking other aspects of the Neumanns but that is what prompted me to move back to the LS50s.

I actually did get close to the Q6 when I was listening to them and when you're close to them it's much easier to tell that some male vocals are coming from the woofers, which isn't surprising since you need distance for the sound fields to converge. At normal listening distances it's much harder to tell but it's still there if you're actively trying to listen for it.

Even if listening in the nearfield would work, that's not how I listen so it wouldn't be a long term solution. It also doesn't make sense that reflections would be causing this effect since any early reflections are going to sound like they're coming from the initial source based on the Precedence effect.

 

[Arindal]

but I'm telling you it's pretty easy to spot once you A/B them, by the way I do that in mono with a single speaker of each so there are fewer reflections and it's easier to spot differences.

I am not disputing your subjective impression, but rather explaining why I think direct sound vertical localization is unlikely to be the cause. Which leaves only indirect localization via reflections or general directivity differences as a potential explanation.

it's pretty easy to spot once you A/B them, by the way I do that in mono with a single speaker of each so there are fewer reflections and it's easier to spot differences.

It is easier to spot differences, but this is not the way the majority of music is mixed and mastered, nor is a mono speaker placement in front of you compatible with stereo. You might notice differences which you cannot spot in a stereo environment as mono is always real source localization and stereo is mostly phantom source localization, so the way our brain perceives reverb, are pretty different.

I bought a pair of Neumann KH120 to compare and noticed the same thing, even though the Neumann are not coaxial, they sounded more like a point source than the R3 in my comparison.

Another indication for reflections or directivity differences in the 500Hz region (or lower) being the cause, as the KH120 is a conventional 2-way inline design with pretty excessive lobing and interference issues as the distance between the drivers is not small and the crossover freq is 1.8K which is according to my experience the most critical range for vertical localization. I have been working on similar speakers for quite a while and exchanged them for Genelec 8341A which was pretty much of an improvement in terms of localization stability and ´point source feeling´.

Even if listening in the nearfield would work, that's not how I listen so it wouldn't be a long term solution

It was not meant as such but as an experiment to narrow down the cause of the localization issues. I always do such experiments when encountering dominant reverb or localization stability issues.

It also doesn't make sense that reflections would be causing this effect since any early reflections are going to sound like they're coming from the initial source based on the Precedence effect.

That is not necessarily the case, early reflections can pretty obviously contribute to localization instability/blur, or in case they come with interference lobing, contribute to general heterogeneity of the image,

 

[Arindal]

, the LS50 Meta as I have shown has smoothly increasing directivity (index) over the whole audio frequency and thus cannot be compared to some local deviations and kinks.

I found broad-band, expressive level differences of several decibels in a continuously decreasing manner to be much more colocating the reverb then minor deviations and kinks w/o dramatic level differences between octave-brand bands.

your extrapolation from those about that the DI in those should be flat is nothing found in literature.

Yes it is. Literature is full of that. Read Linkwitz´ comments on speaker design, like:

"The best loudspeakers ...
Next to a flat on-axis frequency response in free-space, a frequency-independent off-axis or constant power response is the most important loudspeaker parameter. If a loudspeaker is directional, then it should be directional in the same way for all frequencies and at least in the horizontal plane. "

See Link to Linkwitz on general speaker properties

In that year (2009) he also wrote and published an AES paper on the matter of of optimum radiation pattern of a loudspeaker for stereo reproduction. The importance of sticking to these goals becomes clear the moment you look at the effort to achieve constant directivity, as the whole thing with dipoles and 4-way and active correction was solely created for that aim (not saying that this would be my preferred solution to achieve it, but it shows how seriously he took it).

The directivity of the low frequency range of a single coindicential driver is just dictated by the baffle size.

No, that would only be the case with flat baffles and true pointsources. The moment you have curved baffles, non-pistonic or several drivers as well as sound sources being not in phase (dipole, cardioid), the picture changes dramatically.

But the LS50 isn't a big compromise in directivity form and smootheness

I think it is. The directivity index is increasing at what I find to be unacceptable rates between 500Hz and 8,000Hz, particularly driving up the difference in indirect sound level between the two most important frequency bands 0.8-2k vs. 3-5k by a relative +4dB or so. Most of conventional hi-fi speakers being criticized for uneven directivity or directivity errors are actually better.

In the end its probably a mix of directivity, single point source and less edge diffraction.

I do not see a reason to worry about edge diffraction from a modern box-shaped KEF speaker, as they all tend to have the radiation angle of the critical frequency bands defined by means of coaxial arrangement and inner tweeter waveguide. The edges might look edgy but the waves from the tweeter kind of never see them at a level that would be necessary to cause edge diffraction.

 

[thewas]

I found broad-band, expressive level differences of several decibels in a continuously decreasing manner to be much more colocating the reverb then minor deviations and kinks w/o dramatic level differences between octave-brand bands.

Which again is just your personal claim and preference, for others it is different.

No, that would only be the case with flat baffles and true pointsources. The moment you have curved baffles, non-pistonic or several drivers as well as sound sources being not in phase (dipole, cardioid), the picture changes dramatically.

I clearly wrote single coincidential driver, also at the size of such a small baffle and frequencies we are talking about (bass region) the curving of the baffles plays practically no role, its role starts at higher frequencies.

I think it is. The directivity index is increasing at what I find to be unacceptable rates between 500Hz and 8,000Hz, particularly driving up the difference in indirect sound level between the two most important frequency bands 0.8-2k vs. 3-5k by a relative +4dB or so. Most of conventional hi-fi speakers being criticized for uneven directivity or directivity errors are actually better.

Again just your anecdotal statement, if you show the spin of a similar sized and priced loudspeaker which you prefer we can analyse both comparatively.

I do not see a reason to worry about edge diffraction from a modern box-shaped KEF speaker, as they all tend to have the radiation angle of the critical frequency bands defined by means of coaxial arrangement and inner tweeter waveguide. The edges might look edgy but the waves from the tweeter kind of never see them at a level that would be necessary to cause edge diffraction.

The Uni-Q waveguide mainly works for the tweeter and the effect is reduced but there, the reason why they introduced the shadow flare for their flat baffled models and one of the reasons why their best loudspeaker series (Blade) has the baffle shape it has.

 

[Arindal]

Which again is just your personal claim and preference,

No, the original hypothesis of constant directivity and even indirect sound response was developed by respected members of the audio engineering community (like Earl Geddes and Siegfried Linkwitz), and I just happened to find it confirmed by a number of controlled tests.

if you show the spin of a similar sized and priced loudspeaker which you prefer we can analyse both comparatively.

Please specify size and price, I will try to find one which I have had in a controlled test environment and for which data is publicly available.

the reason why they introduced the shadow flare for their flat baffled models and one of the reasons why their best loudspeaker series (Blade) has the baffle shape it has.

Shadow flare is the waveguide continuation around the coaxial? I don't see a contradiction here, I was just saying the whole thing is controlling dispersion in the treble region pretty well and bringing particularly the 3K+ bands to such a narrow radiation pattern that I would not worry.

If I am not mistaken, the concept Blade and its shape predates all box-shaped Reference models. My personal opinion is that the Reference models with several woofers make much more sense, as the slim, fully rounded baffle of the concept with side-mounted woofers contributes to less dispersion control in the lower midrange and continouesly increasing directivity index over a vast range, hence more imbalanced indirect response compared to for example a Reference 4c Meta.