Unmeasurable Acoustic Artifacts

[Steve F]

Soundminded raises a fascinating issue at the end of the “Best AR 12” Bookshelf” thread about the audibility/inaudibility of very subtle acoustic artifacts and acoustic circumstances, and I feel it warrants a new topic.

Being a formally-trained musician and having performed in many large-scale ensembles where musicians are in close proximity to acoustic “impediments” like music stands, risers, and the like, I agree that it is very difficult—if not impossible—to ascribe particular, unequivocal audible characteristics of the effect on the instruments’ sound caused by those nearby structures.

However, the presence or absence of environmental structures—even relatively subtle ones—often does have an acoustic impact on the sound we perceive, and in surprising ways.

Here’s a personal example: When I travel by air, I’m usually in the aisle seat. If it’s a long flight, I invariably close my eyes and try to snooze (although I’m not one of those people who can sleep soundly on an airplane).

So my eyes are closed, and I’m not paying specific attention to the comings and goings of those around me. Yet I can always tell when someone walks by my seat in the aisle—and I’m not talking about hearing their footsteps or hearing the rustling of their clothes.

When a person walks by my seat, their presence changes the reverberant characteristics of my immediate, local audible environment. The sound of the plane’s interior is ‘open and reflective,’ then it ‘closes in’ (akin to the sound environment in a coat closet), then it ‘opens up’ again as they pass by. There’s no mistaking it.

I would offer that the many subtle, detail-oriented design choices made by a loudspeaker designer could have similar effects, especially in the near field. (Let’s face it: If someone walks by my seat in 11A, I can tell. That’s near-field. If someone walks by 27A, I can’t tell. That’s the far-field, and the overall energy distribution that determines the plane’s interior acoustics are unchanged by that person at 27A.)

The takeaway is this: If I were designing a speaker, I’d do everything I could to optimize the near-field performance, like correct driver orientation on the baffle, minimal nearby protrusions, etc. etc. To room energy proponents, the far-field response will swamp those near-field characteristics anyway, right? But doing things “correctly” in the near-field won’t harm the far-field performance. So “just in case” the near-field performance is critical, do it right. Good far-field and good near-field performance isn’t necessarily mutually exclusive. The 9/90/91/92 Series was good near- or far-. I think that’s why they were such good speakers.

Steve F.

[soundminded]

Soundminded raises a fascinating issue at the end of the “Best AR 12” Bookshelf” thread about the audibility/inaudibility of very subtle acoustic artifacts and acoustic circumstances, and I feel it warrants a new topic.

Being a formally-trained musician and having performed in many large-scale ensembles where musicians are in close proximity to acoustic “impediments” like music stands, risers, and the like, I agree that it is very difficult—if not impossible—to ascribe particular, unequivocal audible characteristics of the effect on the instruments’ sound caused by those nearby structures.

However, the presence or absence of environmental structures—even relatively subtle ones—often does have an acoustic impact on the sound we perceive, and in surprising ways.

Here’s a personal example: When I travel by air, I’m usually in the aisle seat. If it’s a long flight, I invariably close my eyes and try to snooze (although I’m not one of those people who can sleep soundly on an airplane).

So my eyes are closed, and I’m not paying specific attention to the comings and goings of those around me. Yet I can always tell when someone walks by my seat in the aisle—and I’m not talking about hearing their footsteps or hearing the rustling of their clothes.

When a person walks by my seat, their presence changes the reverberant characteristics of my immediate, local audible environment. The sound of the plane’s interior is ‘open and reflective,’ then it ‘closes in’ (akin to the sound environment in a coat closet), then it ‘opens up’ again as they pass by. There’s no mistaking it.

I would offer that the many subtle, detail-oriented design choices made by a loudspeaker designer could have similar effects, especially in the near field. (Let’s face it: If someone walks by my seat in 11A, I can tell. That’s near-field. If someone walks by 27A, I can’t tell. That’s the far-field, and the overall energy distribution that determines the plane’s interior acoustics are unchanged by that person at 27A.)

The takeaway is this: If I were designing a speaker, I’d do everything I could to optimize the near-field performance, like correct driver orientation on the baffle, minimal nearby protrusions, etc. etc. To room energy proponents, the far-field response will swamp those near-field characteristics anyway, right? But doing things “correctly” in the near-field won’t harm the far-field performance. So “just in case” the near-field performance is critical, do it right. Good far-field and good near-field performance isn’t necessarily mutually exclusive. The 9/90/91/92 Series was good near- or far-. I think that’s why they were such good speakers.

Steve F.

"So my eyes are closed, and I’m not paying specific attention to the comings and goings of those around me. Yet I can always tell when someone walks by my seat in the aisle—and I’m not talking about hearing their footsteps or hearing the rustling of their clothes.

When a person walks by my seat, their presence changes the reverberant characteristics of my immediate, local audible environment. The sound of the plane’s interior is ‘open and reflective,’ then it ‘closes in’ (akin to the sound environment in a coat closet), then it ‘opens up’ again as they pass by. There’s no mistaking it."

That is correct. When someone walks past you as you describe it, one of your ears, the one on the aisle side passes in their acoustic shadow. Their body literally blocks much sound, including ambient noise from reaching one of your ears nearly as efficiently as it had before they got there and over a very wide angle both vertically and horizontally depending on how close they are. This could be 60 to 90 degrees horizontally and 90 to 120 degrees vertically. When this happens, the similarity of sounds between your left and right ear changes drastically and quickly. This similarity is called "Inter-Aural Cross Correlation" or IACC. The brain quickly and continuously computes this effect and the changes from one instant to the next as the shadow passes alerting you to the presence of the passerby. This is one of our built in instinctive strategies that has evolved in us for survival. The directions from which sounds arrive is very critical.

This is very different from a musical instrument or music stand near a musician. That surface may only be a fraction of a degree where the listener is positioned and the phenomenon is reflection. Of far greater importance at a live performance is the role of the stage floor. It is intended to be hard and reflective, to create many early reflections scattered in many directions including at the listener. The loudness of reflections, angles they arrive at the listener at, and their relative spectral content is critical to the degree and nature of how they change perception of sound. In many cases including the small angle and slight increase in amplitude, this change for a speaker like AR3a is probably negligable. To determine this, an AB listening test and measurements of an AR3a as manufactured and marketed and one with a modified cabinet where there are no reflections from the molding or baffleboard would reveal if there are any significant changes.

When you hear a recording of a musical performance, can you tell if a soloist is alone on stage, standing or sitting, has a music stand or other reflective object near him? There are many wild claims by audiophiles which make no sense to me. For example, they claim to be able to hear absolute phase. But if you are standing near a harpist who plucks a single string, can you tell if on the string's first vibration it is moving towards you creating a compression wave or away from you creating a rarifaction wave? Would you be able to tell in a recording? This is of course an unfair question because if you know how a string actually vibrates, you know that different parts of it are moving in both direcions at once since it breaks up into harmonic modes. A better test would be a wind instrument where the first arriving sound is a compression wave...unless it is pointed at the floor and the first arriving sound is a bounce which causes it to change phase 180 degrees...or does it?

[Steve F]

"To determine this, an AB listening test and measurements of an AR3a as manufactured and marketed and one with a modified cabinet where there are no reflections from the molding or baffleboard would reveal if there are any significant changes."

This would be a great test and would put to rest, once and for all, the question/controversy of driver alignment/cabinet protrusions/near-field 'hovoc' etc that everyone has been debating for decades.

The right way to do it, of course, is to use the exact same drivers and then simply add or remove "cabinet molding" and the like with tightly-controlled listening panels in blind listening sessions. If the MF-HF drivers are mounted side-by-side, simply rotate the cabinet so the drivers are vertical and repeat the listening test. That way, the variability--even very slight ones--of driver-to-driver and crossover component differences are removed completely, leaving the cabinet diffraction elements and driver alignment as the ONLY variables.

I have a hunch that this is a test not too many so-called 'audiophiles' would like to see. Just a hunch.

But as I opined in my first post, being a near-field 'agnostic,' if there's no downside to designing for both good near- and far-field response, then that's the way to go.

Steve F.

[genek]

We probably need to make a distinction between "unmeasurable" and "unmeasurable with current standard equipment and methods." Lining the surfaces of a listening space with several thousand transducers and compiling a multimillion point cloud might very well result in being able to identify the effects of something as small and thin as a music stand.

That, of course, would then require us to make a distinction between "unmeasurable," "undetectable" and "measurable and detectable but irrelevant for all practical purposes" (these are actual categories we use in data acquisition in my day job).

BTW, I think Carl mentioned once that he has, or used to have, a cabinet built to 3a dimensions but with no molding and a flat front. Maybe he's already done the test.

[Mexicomike]

But guys...wasn't all this "research" pretty much done by speaker manufacturers 40 years ago? I remember all sorts of discussions re this back in the day, including demonstrations of various designs/applications/explanations. I'm not seeing anything "new" re speaker design in the above discussion.

[RoyC]

BTW, I think Carl mentioned once that he has, or used to have, a cabinet built to 3a dimensions but with no molding and a flat front. Maybe he's already done the test.

I think it was my cabinets you are thinking of, Gene. I had them built in the mid 90's to the same dimensions as the 3a cabinet, including the bracing, but they were mirror imaged. The grille could introduce a "baffle border", not unlike the original 3a molding, but there was no border with the grille removed. I used it for housing AR-3a, AR-3a Limited, and AR-11 guts. In the past 4 years I've used these cabinets for replacement tweeter experimentation for the AR-3a, the last being a planar driver (as seen in the photo below). In all cases an original pair of 3a's were used as a reference for comparison.

Although differences could be heard between various configurations, I never heard differences I could attribute to the presence, or lack thereof, of the AR-3a baffle. When all drivers and crossovers were the same, the difference between the "new" and original AR-3a was not discernible (at any distance). These were subjective comparisons, as no room measurements were taken. I eventually went back to the original AR-3a cabinet because I found it more aesthetically pleasing.

In these discussions we should bear in mind that the day of monolithic structures residing in homes in the name of "sound" is mostly over. The sound source had better be invisible or pleasing to look at. Incremental differences in the sound of big boxes are irrelevant to all but the very few who still like to debate such things. I personally preferred the sound of the NHT 3.3 over any AR speaker I have ever heard, including the AR-9, but never found a way to fit them into my living space...and never lost any sleep over it.

Roy

[Pete B]

Interesting Roy,

Do you still have the reference pair of 3a's or any others, 11s for example?

Just curious since I am sometimes interested in doing comparisons.

Thanks!

[genek]

But guys...wasn't all this "research" pretty much done by speaker manufacturers 40 years ago?

I suspect that most of the pros chose their preferences and went on a long time ago. Which has never stopped the rest of us.

[RoyC]

Interesting Roy,

Do you still have the reference pair of 3a's or any others, 11s for example?

Just curious since I am sometimes interested in doing comparisons.

Thanks!

Hey Pete,

I still have a number of pairs of 3as, but no AR-11's. The cabinets discussed above currently have no drivers in them.

Roy

[tysontom]

"To determine this, an AB listening test and measurements of an AR3a as manufactured and marketed and one with a modified cabinet where there are no reflections from the molding or baffleboard would reveal if there are any significant changes."

This would be a great test and would put to rest, once and for all, the question/controversy of driver alignment/cabinet protrusions/near-field 'hovoc' etc that everyone has been debating for decades.

The right way to do it, of course, is to use the exact same drivers and then simply add or remove "cabinet molding" and the like with tightly-controlled listening panels in blind listening sessions. If the MF-HF drivers are mounted side-by-side, simply rotate the cabinet so the drivers are vertical and repeat the listening test. That way, the variability--even very slight ones--of driver-to-driver and crossover component differences are removed completely, leaving the cabinet diffraction elements and driver alignment as the ONLY variables.

I have a hunch that this is a test not too many so-called 'audiophiles' would like to see. Just a hunch.

But as I opined in my first post, being a near-field 'agnostic,' if there's no downside to designing for both good near- and far-field response, then that's the way to go.

Steve F.

AR actually did this, I think with the AR-3a and many of the other speakers. The speakers were measured with molding and grill panel removed and compared with the same setup with the molding. No surprises. The main difference was a slightly brighter acoustic-power output due to better dispersion due to a less-obstructed baffle and less absorption, but the spectral balance in the reverberant field was basically the same according to what RFA told me some years ago. Again, the early reflections and interference effects that are quite noticeable in the near field are largely swamped back in the reverberant field. Later on, I don't know if AR reported on a comparison of vertical vs. horizontal placement of the ADD 3-ways, but I would think that differences back in the far field are less noticeable here as well.

As you said, the AR9 tower configuration does an excellent job of compromising good near-field and good reverberant-field response.

--Tom Tyson

[Carlspeak]

I think it was my cabinets you are thinking of, Gene. ........ In the past 4 years I've used these cabinets for replacement tweeter experimentation for the AR-3a, the last being a planar driver (as seen in the photo below).

Roy

Hey Roy, that configuration looks familiar! Customer was extremely pleased with the results.

[Steve F]

AR actually did this, I think with the AR-3a and many of the other speakers. The speakers were measured with molding and grill panel removed and compared with the same setup with the molding. No surprises. The main difference was a slightly brighter acoustic-power output due to better dispersion due to a less-obstructed baffle and less absorption, but the spectral balance in the reverberant field was basically the same according to what RFA told me some years ago. Again, the early reflections and interference effects that are quite noticeable in the near field are largely swamped back in the reverberant field. Later on, I don't know if AR reported on a comparison of vertical vs. horizontal placement of the ADD 3-ways, but I would think that differences back in the far field are less noticeable here as well.

As you said, the AR9 tower configuration does an excellent job of compromising good near-field and good reverberant-field response.

--Tom Tyson

No, AR didn’t conduct actual blind-panel comparative listening tests of molding vs. non-molding, or of different driver-alignment conventions. Not to my knowledge anyway, and not reported on or documented in any widely-published literature.

AR in the 60’s through mid-70’s operated on the absolute assumption that the far-field power response completely dominated the listener’s perception of the speaker’s sound. I’m not opening up that subject for debate again; I’m simply making the historically-factual observation that AR felt that way.

So, it must have been measurements and anecdotal listening by various AR personnel that led them to take that ‘far-field dominates’ stance.

Why, then, did AR publish their FR measurements of just the drivers alone, and with all decorative molding removed?

A few possibilities:

1. AR took those measurements to find out exactly what the drivers themselves were doing, to ascertain precisely how good their ‘building blocks’ were. After all, if the drivers themselves were poorly-done, then the chances of a good finished speaker system were about nil.

But the drivers were excellent, so AR proudly showed their data to the outside world.

AR knew that the cabinet molding and the drivers’ interference effects would wreck those nice-looking curves, so someone there made the decision to only publish the individual curves.

(Yes, granted, if you dug real deep and requested the “AR-3a Data Sheet,” there was a full system FR curve, spliced from an outdoor woofer measurement and with the AR amp’s treble control at 2:00. Not as pretty. And if you dug even deeper and read the Allison-Berkowitz paper on Domestic Soundfields, there were more 3a curves. Even messier.)

2. AR was well aware of all of the above, and even though they firmly believed at that time that the far-field dominated the perceived sound, their Sales/Marketing would not allow widely-published ‘messy’ full system FR curves. So I’m not accusing AR of being deceitful; I’m saying AR would not allow the publication of ‘sloppy-looking’ near-field full system FR curves because AR was likely afraid of how they’d be misinterpreted by an under-knowledgeable public.

AR obviously had much the same corporate philosophy with the ADD models. They now recognized that cabinet protrusions were bad, but still didn’t care about side-by-side driver interference. The 10 Pi/11/12 were all still ‘wrong,’ as far as driver placement was concerned.

So what happened between 1977 (the last year of the 10 Pi) and 1978 (the intro of the AR-9) to change AR’s mind about driver alignment? The 9/90/91/92 were all ‘correct.’ But logic and theory tells us that far-field power response will be significantly the same, regardless of driver alignment and cabinet molding, right? After all, those are just near-field factors.

So, this is the big question:

How and why did AR come to the very specific and unequivocal decision that NEAR-field considerations were suddenly important?

They DID come to that decision. But how? Listening tests? Yes, measurements would show horizontal interference between side-by-side drivers in the near-field, but the FAR-field response would be largely unaffected.

We were told that the near field didn’t matter, according to the 3a and 11.

If it was by listening, then AR’s 60’s to mid-70’s contention that the near-field wasn’t an important determinant of a speaker’s perceived sound in a domestic listening environment was wrong.

If AR was right in the mid-60’s, then the AR-9 was disingenuous marketing sleight-of-hand, preying upon the speaker-buying public’s fears of potential speaker weaknesses, when those weaknesses didn’t actually exist.

Which is it? The AR-3a AND the AR-91 BOTH can’t be right, from a driver-alignment/cabinet-protrusion standpoint. Either it matters or it doesn’t.

If the near-field matters (and I say it does, at least to some extent—feel free to jump on my personal opinion), then the 91 is the better speaker vs. the 3a/11, because its near-field response is far better than a 3a’s or 11’s from a cabinet diffraction-driver alignment standpoint, with no sacrifice in far-field response.

So, the question remains—how/why did AR change its design philosophy from near field diffraction and driver alignment not mattering to it becoming a central design tenet?

Steve F.

[Mexicomike]

Then where does the LST fit into all this since it was a totally different design dispersion-wise than any AR speaker before or since? It seems to me that the LST was the last best speaker of the

"traditional" AR design philosophy. And it seems obvious that the thinking at AR changed rather dramatically afterward. If the LST was the best speaker AR knew how to make in the early 70's, the fact that they did not continue that sort of driver arrangement indicates that either they changed their opinion on what "accurate" sound was or they decided that continuing in the LST mode was just too expensive to produce.

[ar_pro]

Ironically, despite their better measurements, the AR-9 and AR-90 have never seemed all that good to me in near-field conditions - their presentation is BIG, and their "correct" driver alignment notwithstanding, if one listens too close-up, they begin to sound like the world's largest pair of headphones.

That said, although I'd most likely choose an AR-91 over a 3a in a near-filed circumstance, it's probably because the 91 wouldn't require a bookshelf, stands, or other support for optimal placement. The vast majority of AR-3a and other shelf-type systems that I've heard in my life were arranged in much less than prime circumstances, and the 91's floor-standing nature gives it more of an edge, I think, than does its particular driver alignment.

I've heard well-amplified stacked AR-3a systems, LST's and single AR-9 and AR-90 systems in large rooms, and they absolutely do not present in the same manner - there's a definite family resemblance, but they *do not* sound the same - the later models open up with a perceived frequency extension and authority that just isn't there in the older speakers.

Perhaps AR as a company was motivated by the expanding size of homes occupied by its demographic!

[genek]

So what happened between 1977 (the last year of the 10 Pi) and 1978 (the intro of the AR-9) to change AR’s mind about driver alignment? The 9/90/91/92 were all ‘correct.’ But logic and theory tells us that far-field power response will be significantly the same, regardless of driver alignment and cabinet molding, right? After all, those are just near-field factors.

So, this is the big question:

How and why did AR come to the very specific and unequivocal decision that NEAR-field considerations were suddenly important?

My guess is that under Teledyne management that came after Roy Allison's departure the company moved from product development based on design philosophy to one based on marketing, and near field considerations became important because it was what the consumer market had come to demand.

[Steve F]

My guess is that under Teledyne management that came after Roy Allison's departure the company moved from product development based on design philosophy to one based on marketing, and near field considerations became important because it was what the consumer market had come to demand.

This answer implies that AR knew all about near-field response and its implications on listener perception all along, and that no new engineering knowledge was acquired along the way during the years from the 3/3a to the 9/91.

My contention is that while Roy Allison's work at AR from the early 60's to 1972 was seminal in nature, it wasn't all-encompassing of EVERY SINGLE acoustic engineering factor for all time, forever. Give Tim Holl and his team their due for discovering what they did when conducting the research that led to the AR-9. The idea of vertical driver alignment and suppression of near-field baffle reflections/cabinet diffraction was important work.

The Allison-era AR simply did not think of this yet. They were more concerned with bigger fundamentals, like getting the drivers correct, discovering the relationship between driver size and dispersion, doing the proper loading of bass drivers into the appropriate cabinet volume, and doing proper elemental crossover design.

Once that had been done, that knowledge existed already for Holl and other designers of his day to build upon.

The thought of some faceless execs at Teledyne sitting around a large conference table and proclaiming, "Thou shalt design for near-field considerations!" strikes one as fanciful.

Instead, it's more likely that the industry's collective knowledge and understanding had simply moved on to the point where near-field considerations--as well as power response, if you still wanted it--played a larger role in the priorities of that day's top designers.

It remains: if the 91 is correct, then the 3a can't be. The 3a can be 'mostly' correct, it can be 'substantially' correct, it can (and does!) sound very, very good, but it can't be 'as correct' as the 91 if Holl's work is regarded as legitimate.

Note: The ENTIRE LOUDSPEAKER INDUSTRY follows Tim Holl's design principals to this day: Vertically-aligned drivers, and minimal cabinet reflections. (True--the ADD's--before Holl--had reduced cabinet diffraction, but not to the extent that his Blanket designs did.) The details of how different companies achieve minimal, non-destructive cabinet reflections varies from company to company, but all companies now regard that and vertical alignment as a given. The execs at Teledyne didn't mandate that!

Steve F.

[genek]

This answer implies that AR knew all about near-field response and its implications on listener perception all along, and that no new engineering knowledge was acquired along the way during the years from the 3/3a to the 9/91.

Not necessarily. If they had previously operated on the belief that far-field performance dominated the listener experience, they might not have made much of an effort to investigate near-field performance beyond what was necessary to satisfy them that it had no significant effect on the far field. A marketing conclusion that speaker consumers were making near-field performance a deciding factor in their purchasing decisions would have been the impetus for them to start developing more near-field engineering knowledge.

I would guess that the marketing decision was less of a "thou shalt" and more of a "maybe we need to." It's not as if incorporating newer near-field tech did anything to degrade the far-field performance that the previous generation's designers had concluded was unaffected by it anyway.

And the fact that the entire industry now follows the near-field design philosophy probably will not change the opinion of some that it is all silliness because the near-field effects are swamped in the far field.

[speaker dave]

So, this is the big question:

How and why did AR come to the very specific and unequivocal decision that NEAR-field considerations were suddenly important?

They DID come to that decision. But how? Listening tests? Yes, measurements would show horizontal interference between side-by-side drivers in the near-field, but the FAR-field response would be largely unaffected.

So, the question remainshow/why did AR change its design philosophy from near field diffraction and driver alignment not mattering to it becoming a central design tenet?

Steve F.

It seems to me that this was simply part of the evolution of loudspeaker design. It was driven by evolving design capibilities, by competition from other designers, and by expanded knowledge of what was important.

Evolving design capibilities went roughly like this: In the 40s and early 50s you might design a network from textbook formulas. Few had the ability to measure the acoustical result which would likely be poor. Later you might take in-room curves or use a reverb chamber to hit a far field or power response shape with no control over response on any particular axis. The third step would be to achieve in-phase summing of drivers on a reference axis. Doing this easily required the ability to measure phase response. The final step would be full network modeling. This would allow "acoustic Butterworth" or other direct field centered approaches. Note that none of these more sophisticated approaches to designing for good near-field performance will neccessarily degrade the far field performance.

A lot of the design competition came from the UK in the mid 70s to mid 80s. Read the articles of Harwood about the BBC designs, plus look at the techniques that KEF, B&W and AR's UK side were developing. The Canadian side also became influential a little after the UK dominance. An ability to create a system with a very linear direct field became commonplace.

Read the papers of Toole, where he shows that flat and smooth free field response is a strong determinant of rank in controlled listening tests (power response is a poor determinant). Kates, Bech and Salmi show the audibility of early room reflections. Lipshitz and Vanderkooy show that holes in the power response are inaudible. Studies of PA systems showed that the direct response still dominated perceived balance when it was as much as 15dB below the reverberent level. All point to the direct sound being more important than the reverberent field.

We design systems with a much flatter and smoother direct field partly because we can (and couldn't 40 years ago), but mostly because we think it is the best way to go.

David S.