Jump to content

The Goals for an "Ideal Loudspeaker"


Zilch

Recommended Posts

Lecture by Siegfried Linkwitz at NorCal DIY meeting today, April 24, 2010, Roseville, CA.

His website: http://www.linkwitzlab.com/

Key points:

1) "The room is usually considered to be the problem when a loudspeaker does not sound right. Actually, the loudspeaker is the problem, because it illuminates the room unevenly with sound at different frequencies."

2) The highest priority in loudspeaker design should be accorded to controlled (constant) directivity.

3) "The ideal loudspeaker is room-independent."

["De-reverberation" time for localization: >6ms.]

Publication #30 here:

http://www.linkwitzlab.com/publications.htm

Link to comment
Share on other sites

  • Replies 174
  • Created
  • Last Reply

Hey Zilch,

I know you are a fan of constant directivity horns and waveguides and you know my history with them. Here are a few questions that might get the conversation going. These pertain somewhat to what Linkwitz is advocating.

What is the goal with constant directivity drivers in loudspeakers? Is it about flat power, or flat power over a part of the frequency range? Is the objective that all wall reflections have the same frequency response?

Assuming we can agree that constant directivity is a good attribute, does that apply to any amount of directivity? What is the optimum d.i. or beamwidth? Must the woofer match the d.i. of the upper range?

Will constant directivity give us better power response? Is it essential for optimum power response?

What is the right trade off between power response and direct response?

David

Link to comment
Share on other sites

It's interesting to follow Linkwitz's learning curve in his lectures over time as linked on his publications page; the focus on constant directivity is recent. His theme on Saturday, as I interpreted it, at least, was that the power response should be uniform such that the early reflections do not "confuse" the spatial accuracy. He did comment that most of the DIY designs demoed at the meet would not have the requisite directivity to accomplish this.

He emphasized symmetry as part of the formula, but also appears to recognize that his dipole designs effectively attenuate the earliest ipsilateral (nearest sidewall) reflection. Asked (by another) in the Q&A whether the Geddes extreme toe-in alignment (described years earlier in your 4430 papers) would accomplish the same using waveguides, his response was, "Yes, very well."

In researching his site, I found the illustration of the reflective condition in small rooms below. Note particularly the angles operative in the phantom sources; the near-wall one originates from well off-axis with modest toe-in, and the other very early forward ones from the rear of the speaker:

http://www.linkwitzlab.com/stereo%20reproduction.htm

post-102716-1272316453.jpg

Link to comment
Share on other sites

On this matter, clarification of Geddes's position with respect to delay:

The first 10 ms is the critical time for localization then out to 20 ms coloration (Tembre) is affected. Beyond 20 ms there is no effect at all on localization, but coloration can still be a factor if the loudspeakers don't have matching power and direct response curves. If they are CD then beyond 20 ms reflections are only a good thing - they create spaciousness - especially if they are lateral.

http://www.diyaudio.com/forums/multi-way/1...tml#post1506550

Apparently affirming agreement with Geddes in his response to another question from the audience, Linkwitz suggested that LEDE should be reversed in typical listening rooms, with the live end behind the listener(s) enhancing spaciousness....

Link to comment
Share on other sites

I think SL needs to take the next step beyond his theory and measurements and verify all of his claims regarding the speakers being the 'problem' and not the room by conducting rigorous listener panel listening tests - after T&O. He's a great engineer, but never-the-less, people have listening preferences embedded in their heads which need to be factored into a sample size which will negate any bias on the part SL or any others and provide a true perspective on what folks like and dislike with respect to C.D., D.I. reverb time (ms), reflections, etc, etc., etc.......

Real people from different walks of life who love music need to listen to his 'ideal' speaker and others that perhaps disperse the sound more, like LST or Allison designs.

Link to comment
Share on other sites

In researching his site, I found the illustration of the reflective condition in small rooms below. Note particularly the angles operative in the phantom sources; the near-wall one originates from well off-axis with modest toe-in, and the other very early forward ones from the rear of the speaker:

Yes, I can see that the sidewall reflection looks to be from a pretty extreme radiation angle, between 45 and 60 degrees.

I think Linkwitz is a brilliant engineer (I love his sketches). I think one of his recent kicks is flat spectrum for the reflections, hence his interest in constant directivity. I agree with Carl that listening tests would be needed to verify the benefit of this. Others have proposed this in the past, notably Queen, but the rational usually goes something like "if the reflections match the direct sound they can't screw up the balance, right?" That isn't the same as proving that it is a necessary condition for the ideal speaker.

The Lipshitz and Vanderkooy study and numerous papers of Toole's show that significant holes in the power response (i.e. leading to reflections with significant dips) don't prevent a speaker from being highly ranked in a listening test.

It gets back to your expectations with constant directivity drivers. Which particular possible benefit is important? You don't really get flat d.i., except over the range of the particular driver. You can have smoother power response, but I haven't seen any recent credible papers that advocate that. On the other hand, I think a very even frequency response over a useful range of possible listening angles is a real benefit. It would allow the high toe-in for time-intensity trading and also give real latitude for listening position.

David

Link to comment
Share on other sites

The "Ideal" loudspeaker is an abstract design construct; I don't believe Linkwitz claims to have built one.

[His Orions sounded mighty good there, tho. :) ]

Yes, I can see that the sidewall reflection looks to be from a pretty extreme radiation angle, between 45 and 60 degrees.

Though I have calculated it previously, this is the first time I've seen that so clearly illustrated.

Given that it has the greatest potential for adulterating the program presentation, I have yet to see a convincing argument that it's important to maintain it when it's so easily mitigated instead. From all of the studies, the later low-IACC contralateral reflection can provide the requisite ASW spaciousness cues without messing up the direct field.

An interesting perspective from another poster in that diyAudio thread: "Stereo is the first reflections with the direct source MIA...." :(

In small rooms these small time differences make BIG perceptual differences because the time delays are always in this range.

The "Summing localization" and "precedence" effects are widely overstated in these discussions because they indicate - as you said - the principle direction of the sound perception. They do not indicate whether there is an increase in image blur or coloration, both of which exist for these small delays. SO yes the "principle" direction is set very early 1-2 ms, but the stability and coloration of that image is strongly influenced by the next 8-10 ms. I want a stable and uncolored image, not just a "good idea" of which direction its in. Blauert admits such in his "Spatial Hearing" book.

He's pretty unequivocal about VER (Very Early Reflections) <10ms and getting rid of them as a necessary condition for an ideal loudspeaker:

Reflections after 15 ms are highly desirable, especially if lateral, because they create spaciosness. Delays less than 10 ms do no good at any time. It is not easy to lower all reflections < 10 ms. by AT LEAST 5 dB (I'd say 10 dB) unless you make the room very dead, but then you kill the spaciousness. You must use high directivity speakers pointed so that they minimize the early reflections, but still excite the later reflections by using a lively room. Wide directivity speakers cannot achieve the low reflections < 10 ms without having a very dead room.

This is precisely where the speaker and room have to work together to achieve the right balance of direct to refections delay and spectral balance between the direct sound and the reverberant sound. Only a CD device of narrow directivity can do this. Nothing else works. And even with narrow directivity, you still almost always have a floor and ceiling reflection that you must do something about (because most rooms are shorter than they are wide). Vertical reflections are never a good thing, although if > 20 ms they are kind of benign especially if not from the forward direction.

There doesn't have to be only the two extremes - modify the room to fit the loudspeaker OR modify the loudspeaker to fit the room. The ideal is to find what combination of both yields the best result. I don't think that the optimum can be reached with only doing the room OR the loudspeaker - you must consider both and work them together as a single system to be optimized.
Link to comment
Share on other sites

Hey Zilch,

I know you are a fan of constant directivity horns and waveguides and you know my history with them.

David

Hi David,

You mention your history with them and I'm wondering where I can read more about your work

if it was published. I remember a paper with Keele in the AES, is this the main one or are there others?

http://www.aes.org/e-lib/browse.cfm?elib=11970

This one is in the yellow Anthology by the way, if anyone is interested.

Here it is at Harman in .pdf format:

http://www.harman.com/EN-US/OurCompany/Tec...ations/4573.pdf

I also found this through google:

http://www.audioheritage.org/html/profiles/jbl/4430-35.htm

I'd just like to learn more about your views on the subject.

Link to comment
Share on other sites

Hi David,

You mention your history with them and I'm wondering where I can read more about your work

if it was published. I remember a paper with Keele in the AES, is this the main one or are there others?

I'd just like to learn more about your views on the subject.

Hi Pete,

Between the AES paper and the Lansing Heritage recollections, that is all that I've written on the JBL systems. You should also read Don Keele's papers on the horn designs at EV and JBL, plus the Altec Mantaray paper.

One caveat is that I think these were a real improvement in high level studio monitors. A bad horn can be very bad indeed and the Keele CD biradials were amazingly good. I'm not sure I would recomend them as the ultimate domestic solution, though. If you don't need the extra 15dB of output then the expense and compromises stemming from compression drivers and horns are a poor compromise for most audio enthusiasts (there, I said it).

What I still find to be important is that the constant directivity flare can be adapted to a tweeter baffle, as Genelec and others have done. Do it right and you can get smoother response, less variation with a wide range of angles, and a little flatter directivity better matching the woofer's. These are, to me, all part of the ideal speaker.

I was looking through Toole's papers and book again. The reason I keep falling back on that is that it is the only source of methodically done blind listening tests, with ranking, and corresponding multiple measurements. How else would we ever discover the link between what we like and how it measures?

It is very clear when looking at the progression of ranking from poor to good, that the best ranked speakers have the widest, flattest and smoothest on axis frequency response. They all have falling power response and no particular shape to the power response (holes in the power response always exist and are placed according to whether the system is 2 way or 3 way. It doesn't appear to effect the ranking.) It is not clear that high or low directivity is an asset. There is no evidence that flat power would be a good thing, even over part of the range.

Is that all we need to know about the ideal loudspeaker?

David

Link to comment
Share on other sites

The "classic" LEDE configuration already has the live end behind the listener. The padded end (dead end) is up front.

Well, maybe they DON'T agree on this point, then, which is starting to make more sense to me, now.... :(

Link to comment
Share on other sites

I was looking through Toole's papers and book again. The reason I keep falling back on that is that it is the only source of methodically done blind listening tests, with ranking, and corresponding multiple measurements. How else would we ever discover the link between what we like and how it measures?

Me too, actually, in formulating a response to the questions you raise at #2 and #6 here.

Unless you or someone else knows better, it's a valid point that the listening test conditions for much if not all of this work did not include an acoustically correct small room, "correct" meaning replicating typical home listening spaces. That has changed since 2007 with the construction of Harman's International Reference Room with in-wall speaker shuffler.

There is, however, a paper in which the same group of speakers was evaluated in different rooms, which I am having difficulty locating. I believe Carl linked to it in one forum or another. Anybody have a clue?

Link to comment
Share on other sites

Me too, actually, in formulating a response to the questions you raise at #2 and #6 here.

Unless you or someone else knows better, it's a valid point that the listening test conditions for much if not all of this work did not include an acoustically correct small room, "correct" meaning replicating typical home listening spaces. That has changed since 2007 with the construction of Harman's International Reference Room with in-wall speaker shuffler.

There is, however, a paper in which the same group of speakers was evaluated in different rooms, which I am having difficulty locating. I believe Carl linked to it in one forum or another. Anybody have a clue?

Is this it?

http://seanolive.blogspot.com/2010_03_01_archive.html

Link to comment
Share on other sites

Not the one, but certainly interesting.

Most anyone with experience doing loudspeaker measurements can tell from just a couple seconds' burst of broadband noise whether the frequency response will pass muster, before the curve has even appeared on the screen.... :(

Link to comment
Share on other sites

Hi Pete,

Between the AES paper and the Lansing Heritage recollections, that is all that I've written on the JBL systems. You should also read Don Keele's papers on the horn designs at EV and JBL, plus the Altec Mantaray paper.

One caveat is that I think these were a real improvement in high level studio monitors. A bad horn can be very bad indeed and the Keele CD biradials were amazingly good. I'm not sure I would recomend them as the ultimate domestic solution, though. If you don't need the extra 15dB of output then the expense and compromises stemming from compression drivers and horns are a poor compromise for most audio enthusiasts (there, I said it).

What I still find to be important is that the constant directivity flare can be adapted to a tweeter baffle, as Genelec and others have done. Do it right and you can get smoother response, less variation with a wide range of angles, and a little flatter directivity better matching the woofer's. These are, to me, all part of the ideal speaker.

I was looking through Toole's papers and book again. The reason I keep falling back on that is that it is the only source of methodically done blind listening tests, with ranking, and corresponding multiple measurements. How else would we ever discover the link between what we like and how it measures?

It is very clear when looking at the progression of ranking from poor to good, that the best ranked speakers have the widest, flattest and smoothest on axis frequency response. They all have falling power response and no particular shape to the power response (holes in the power response always exist and are placed according to whether the system is 2 way or 3 way. It doesn't appear to effect the ranking.) It is not clear that high or low directivity is an asset. There is no evidence that flat power would be a good thing, even over part of the range.

Is that all we need to know about the ideal loudspeaker?

David

I'm finding my thoughts to be completely in line with yours David and I was not aware of your work on the JBL horn based monitors so that's why I am interested. I do often mostly agree with Linkwitz, but some of his recent statements of "fact" seem to have no experimental basis or confirmation as you too have pointed out and I was about to write here that Harmon is probably the only company with the resources to reliably test such theories - but you beat me to the statement. I never completely agreed with the dipole bass is the only way theory and it bothers me that it is so wasteful of volume displacement however he does obviously do a completely competent job designing them.

As far as the question of the ideal loudspeaker, I'd have to state that from my perspective it depends. It depends on too many variables to try to pretend that we could come up with the ideal loudspeaker. It depends on the size of the room, if it is treated at all, and many other factors. It is a good thought exercise I suppose.

I'd like to back up here and ask a simpler question (perhaps I should start another thread) concerning something that I noticed years ago. There are speakers that clearly sound like music coming out of a box, for me Advents and ARs were in the long list of speakers with this quality years ago when I first heard them, Dynaco A-25s and EPI100s less so back in the day. Then there are others, often audiophile speakers that disappear, one - they do not sound like music from a woofer and tweeter (not well integrated) and even better the speaker seems to vanish - no box, perhaps close enough reproduction to trick the ear-brain into thinking that the music is here in the room. For me, PSB Stratus Golds are a good example of a speaker that tests well and also pulls off the disappearing act well, and has outstanding dynamics. There is a distinct boxy quality from many speakers and I'm fairly certain I know what causes this. There are also speakers that as a stereo pair do a good job of reproducing a center image, a solo performer for example, and I believe that there are some tricks to pull this off also. I am surprised that this is never mentioned in all the work done at Harman. I bring this up because Linkwitz is finally talking about it but I disagree with his theory as to what is required to attain it:

"It is possible to reproduce a stereo recording in an ordinary living room such that listeners have the illusion that the two loudspeakers have disappeared. When they close their eyes, they can easily imagine to be present at the recording space, as they listen to the phantom audio scene in front of them.

The vast majority of loudspeakers that have been sold - the typical box speakers - can only produce this effect to a limited degree because of a fundamental limitation: they radiate sound into the room with different intensity at different frequencies and angles, though they measure flat on axis. Consequently the many reflections from room surfaces are sonically colored in a way that is characteristic for box loudspeakers. We always recognize the sound as coming from a box rather than being live. It is the generic loudspeaker sound."

Link to comment
Share on other sites

Not the one, but certainly interesting.

You might be thinking of Toole's part 2 paper of his listener preference papers. There was a lot on rooms, measurements and how to take them in those papers.

Link to comment
Share on other sites

As far as the question of the ideal loudspeaker, I'd have to state that from my perspective it depends. It depends on too many variables to try to pretend that we could come up with the ideal loudspeaker. It depends on the size of the room, if it is treated at all, and many other factors. It is a good thought exercise I suppose.

We're taking the room out of the picture. The ideal loudspeaker doesn't care whether it's in a good one or bad one.

[David hasn't weighed in on this goal yet. :( ]

Edit: The paper is:

S. E. Olive, P. L. Schuck, S. L. Sally, and M. E.

Bonneville, “The Variability of Loudspeaker Sound Quality

among Four Domestic-Sized Rooms,” presented at the

99th Convention of the Audio Engineering Society, J. Audio

Eng. Soc. (Abstracts), vol. 43, pp. 1088, 1089 (1995

Dec.), preprint 4092.

See §6. Adaptation, here:

http://www.harman.com/EN-US/OurCompany/Tec...tions/13686.pdf

Spot on point.... :)

Link to comment
Share on other sites

We're taking the room out of the picture. The ideal loudspeaker doesn't care whether it's in a good one or bad one.

How is this done? I'm still not seeing how the ideal speaker can not care about what kind of a room it's in when it's being heard by non-ideal listeners in non-ideal rooms.

Do ideal speakers come with a couple of hundred pounds of anechoic lining you apply to the walls, floor and ceiling of your living room...?

Maybe if someone can figure out a way to make speakers with distance-limited output and a pattern similar to that of a cardiod microphone so the sound field never reaches any room surfaces...?

Link to comment
Share on other sites

How is this done? I'm still not seeing how the ideal speaker can not care about what kind of a room it's in when it's being heard by non-ideal listeners in non-ideal rooms.

Read that link. The room's not such a big deal as popular wisdom would have us believe.

It's easily dealt with, actually, but that's off-topic; we're in goal-setting mode here.

[Also of note: we don't hear differently, either.... :( ]

post-102716-1272431342.jpg

Link to comment
Share on other sites

We're taking the room out of the picture. The ideal loudspeaker doesn't care whether it's in a good one or bad one.

[David hasn't weighed in on this goal yet. :( ]

Edit: The paper is:

S. E. Olive, P. L. Schuck, S. L. Sally, and M. E.

Bonneville, “The Variability of Loudspeaker Sound Quality

among Four Domestic-Sized Rooms,” presented at the

99th Convention of the Audio Engineering Society, J. Audio

Eng. Soc. (Abstracts), vol. 43, pp. 1088, 1089 (1995

Dec.), preprint 4092.

See §6. Adaptation, here:

http://www.harman.com/EN-US/OurCompany/Tec...tions/13686.pdf

Spot on point.... :)

Adaptation is a fascinating subject and I have not seen a lot written about it. I noticed the effect myself however it was not fast, not moment to moment.

I often find in store show rooms, or at shows that when I first walk into the room the sound is quite bad to me, then if I leave and come back a while

later it seems that perhaps some adaptation has taken place. But I believe that the adaptation is more about perhaps "de-reverberation" as Linkwitz

puts it - there has to be a better term for that by the way, rather than frequency response correction. Although, I have seen FR adaptation in other

situations. I am certain that there is a limit to this adaptation since for example, I find that I am unable to adapt to a large room mode bass resonance

problem, or a highly reverberant room. Very early reflections that cause comb filtering are also an issue as I see it.

It is complex and these are just casual observations of mine.

Do you know of any more papers on adaptation?

Let me just summarize by saying that I think it is absurd to claim that we are able to adapt to any room, even most home listening rooms especially if

you allow the speakers to be placed where people want them rather than the ideal locations. Bass management/correction and multiple sub

solutions are further evidence that full adaptation is not possible.

Link to comment
Share on other sites

The room's not such a big deal as popular wisdom would have us believe.

I've wondered about that, because I've had my speakers in everything from a single-room apartment to a 1500 sq ft great room and I'm pretty sure that what has defined the listening experience most for me is the relative width of their direct output rather than room reflections. The different rooms the speakers have been in over the years seemed to affect bass more than anything else.

Link to comment
Share on other sites

I have been re-reading Sean Olive's papers and blog lately and while I admire the work that

they are doing at Harman, I have to wonder if their training process introduces bias into

the listeners. If they train them to prefer a speaker that measures flat, even if what they

hear does not have a natural tonal balance then they are introducing bias. To put this

another way, they should be comparing/training against a live performance not what is in

their view technically correct. I'm not certain how they actually do it, but I do believe that

there was a paper on the subject of listener training that might cover the details.

Link to comment
Share on other sites

Bass management/correction and multiple sub solutions are further evidence that full adaptation is not possible.

Bass is modal, HF, statistical. You can't habituate to what you can't hear.

[Other than to not hearing it.... :( ]

Link to comment
Share on other sites

Bass is modal, HF, statistical. You can't habituate to what you can't hear.

[Other than to not hearing it.... :( ]

You state the obvious; I can certainly hear a modal peak as I already stated.

It would help if you used a few more words to explain yourself.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...