The Insanity of Marketing Disguised as Science in Loudspeakers

[haraldo]

Moving away from center means more distance from one speaker, less distance to the other. Than means that you can easily go from 0/0 to +6db/-6db with just a few feet of movement. It doesn't matter what your speaker does, you can't change that.

With a line source speaker the effect of this is being halved, compared to a point source....

For a line souce: For every distance doubled sound pressure goes down 3dB
For a point souce: For every distance doubled sound pressure goes down 6dB

Does this also impact off axis response?

 

[Nuance AH]

If I'm rude... and I am sometimes, then I believe many loudspeaker manufacturers throws the phase coherence out the window because it's too difficult for them to deal with....

One problem of course with mostly all products using shallow slope crossovers is the load on the drivers so dynamics may suffer....

I partially disagree with the first comment, but do agree with the second. I think the two main reasons manufacturers throw time and phase accuracy out the window is because:

1) Proving it makes an audible difference has been difficult up through present day, so why bother utilizing it.
2) The trade-offs (dynamics/sensitivity, off-axis performance, etc) aren't worth adding an attribute that may or may not actually make an audible difference.

With that said, there is something special about the Model 5A and 7, though I cannot directly relate it to time and phase accuracy. I think it's just darn good speaker engineering with a "price no limit" mentality.

 

[MinusTheBear]

I partially disagree with the first comment, but do agree with the second. I think the two main reasons manufacturers throw time and phase accuracy out the window is because:

1) Proving it makes an audible difference has been difficult up through present day, so why bother utilizing it.
2) The trade-offs (dynamics/sensitivity, off-axis performance, etc) aren't worth adding an attribute that may or may not actually make an audible difference.

With that said, there is something special about the Model 5A and 7, though I cannot directly relate it to time and phase accuracy. I think it's just darn good speaker engineering with a "price no limit" mentality.

Kevin Voecks of Revel has stated before that John Vanderkooy and Stanley Lipshitz at the University of Waterloo in Canada looked at the audibility of time coherence with a specialty built box by allowing you to alter phase response without altering the amplitude.

 

[haraldo]

Kevin Voecks of Revel has stated before that John Vanderkooy and Stanley Lipshitz at the University of Waterloo in Canada looked at the audibility of time coherence with a specialty built box by allowing you to alter phase response without altering the amplitude.

Reading the "Audio Perfectionist Journal" by Richard Hardesty there's lots of credible proof that states otherwise.... it's possible to throw handgranates at each other about this but some people will never agree.

Even Michael Fraemer in Stereophile "came out of the closet" suggesting the Vandersteen Quatto are clearly superior to his super high end Wilson speakers that are many times the price of the vandy's......

 

[Nuance AH]

Kevin Voecks of Revel has stated before that John Vanderkooy and Stanley Lipshitz at the University of Waterloo in Canada looked at the audibility of time coherence with a specialty built box by allowing you to alter phase response without altering the amplitude.

What were their findings? Those were the same gentleman that tested the audibility of cabinet resonance and the effects of bracing/cabinet thickness. They were some busy boys back in the day, and I am grateful for their hard work.

Even Michael Fraemer in Stereophile "came out of the closet" suggesting the Vandersteen Quatto are clearly superior to his super high end Wilson speakers that are many times the price of the vandy's......

That only proves the Wilson's suck, like I've always said they do.

Reading the "Audio Perfectionist Journal" by Richard Hardesty there's lots of credible proof that states otherwise.... it's possible to throw handgranates at each other about this but some people will never agree.

Not that I disagree with you, but Richard is only one guy, and a known time and phase accuracy advocate. I think he also has a love affair with Vandersteen speakers. I'd be curious to know John Vanderkooy and Stanley Lipshitz's findings.

 

[haraldo]

Not that I disagree with you, but Richard is only one guy, and a known time and phase accuracy advocate. I think he also has a love affair with Vandersteen speakers. I'd be curious to know John Vanderkooy and Stanley Lipshitz's findings.

Yes he's a very outspoken advocate of this but after selling all kinds of brands of speakers that you can mention and think about, so if he's objective..... I don't know..... but to me it's credible... I can't show the papers here or links to the because they're licensed so you have to buy them......

His most outspoken advise though is to give yourself time and listen.... and listen.... and listen.... even for month's before you choose a speaker....

Also Mr Hardesty is a very clear adviser towards trying to look beyond the flashy lights and the hype in the ads....

I'd also be curious to know John Vanderkooy and Stanley Lipshitz's findings.....

 

[gene]

I'm aware of only a few manufacturers who publish papers with DBT results -- Harman is one, Pioneer's Japanese research arm another that come to mind off the top of my head. So, will you tell us just who you are referring to when you claim, for example, that manufacturers' published DBTs suffer from 'familiarity bias' and that the researchers have not considered this? Can we lay that claim to rest about Harman at least? If you've actually read Toole's and Olive's papers you know they have brought in outside listeners for their DBTs, as well as used Harman employees. (And much of the early Toole paper work was done for the Canadian NRC, not a manufacturer)

I will say Harman is probably one of the few companies that don't suffer from "familiarity bias" in their test results. Yes I have read their test results and I still feel their sample size (alternative brands / models) are too small and limited to come to some of the conclusions they have come to so definitively. Harman only directly compares their products to whom they feel are their mainstream direct competitors (ie. B&W, Martin Logan,Klipsch, etc). There are over 400 brands in the US market alone, some of whom don't have the marketshare and thus fall off Harman's radar. There are a lot of good companies doing good work making expensive but excellent product. Thus I disagree with the notion beyond a certain $$$ only cosmetic improvements are achieved. I don't think we can define such an amount when our ability to make subjective comparisons and conclusions are still quite limited. Most listening tests are done at low power levels where speakers generally behave well. What happens at higher levels in large rooms? Clean output and extension costs $$$s b/c they involve multi driver arrays, more powerful motors, better high temp crossover parts, etc. $1500/pair is NOT the magic # for achieving this (not saying Harman says this, but some brands that preach DBT as religion do).

Upon closer inspection of some tests I've seen, they often seem more like a test of preference towards loudspeakers with more bass extension than overall fidelity. In a direct instantaneous comparison, I've personally found listeners gravitate towards the speaker with most boom and sizzle. This is why I maintain long term and frequent listening tests of products on an individual basis is necessary. If Speaker A has more bass than Speaker B, but Speaker B produces better mids/highs and can play at higher SPL more cleanly, than it may be a good idea to bass manage both speakers to a level matched sub and run separate listening tests to see how listener preference changes. Nobody does this to my knowledge.

IMO we are NOT at the stage (perhaps we never will be) where we can just look at a graph, flip a switch for a quick test and determine causation of measured results vs subjective preferences. Sorry, there are too many variables to control, some of which we still don't have a firm grasp of. Harman is probably closer to this than anyone, but its still not an exact science IMO.

 

[MinusTheBear]

What were their findings? Those were the same gentleman that tested the audibility of cabinet resonance and the effects of bracing/cabinet thickness. They were some busy boys back in the day, and I am grateful for their hard work.



That only proves the Wilson's suck, like I've always said they do.


Not that I disagree with you, but Richard is only one guy, and a known time and phase accuracy advocate. I think he also has a love affair with Vandersteen speakers. I'd be curious to know John Vanderkooy and Stanley Lipshitz's findings.

Those findings were that phase errors are audible in anechoic chambers and with headphones not in actual reflective listening environments.

 

[GranteedEV]

Not that I disagree with you, but Richard is only one guy, and a known time and phase accuracy advocate. I think he also has a love affair with Vandersteen speakers. I'd be curious to know John Vanderkooy and Stanley Lipshitz's findings.

If you've got access to it, the paper is called ""On the Audibility of midrange phase distortion .....(something)....". My university probably lets me access this stuff but i've never really tried. Here's a good substitute, though:

http://www.audioholics.com/education/acoustics-principles/human-hearing-phase-distortion-audibility-part-2

the step response replaces a bunchload of measurements, but not all of them

Right, but I overall don't pay much attention to it. My point was, on the heirarchy of important stuff, where does step response sit?

OK, there must be some sacrifices and if you want phase/time coherence you probably have to at least sacrifice on maximum SPL, I'm not so sure about off axis, polar response.... you probably know much better than me....

As far as vertical off-axis response, any time you have a crossover(IE two drivers operating in the same frequency band with Center to Center spacing larger than 1/4 wavelength), you've sacrificed some. The more overlap in the crossover, the more sacrifice in vertical off-axis response. A 6db/octave crossover will have the least phase distortion, but the most overlap.

As far as horizontal off-axis response (and this applies to vertical as well), it depends on the width of a driver operating at a given frequency.

Even a 4" driver has narrowing dispersion at 2khz. So a shallow slope on it will make its narrowing dispersion play up higher. But also, to get a shallow slope on a tweeter, you'll need 2X, and more likely 3x the crossover point. So not only are you making the 4" driver play higher due to a shallow slope, you're rolling it off much higher. You're probably asking a 4" driver to play up to ~5-6khz as its -3db point, and much higher on a whole. And to add to that yet again, the higher crossover point sacrifices center to center spacing, which is another blow to the vertical off-axis response.

And then add that the 4" driver is playing much lower because it too, needs a shallow slope. Chances are in a 3-way (and more likely 4-way or even 5-way), there's a lot of bandpass gain. So actually, with shallow slopes necessary for phase coherence, you have a lot of drivers contributing in the same frequency region. So at 2khz, it's not just a 4 inch mid, and a 1" tweeter that's 8-10db down, but there's also a lot of bleeding in of the midbass driver, which you can repeat all of the above with respect to off-axis response. Let's say it's an 8" midrange driver crossed over at 500hz. That's also only around 12db down at 2khz. It's contributing, however slightly, to more on-axis response than off-axis response on a whole. Now add one more factor in. An 8" midbass driver that's only 12db down at 2khz, needs to not be breaking up. So you probably have to use a softer cone material that's probably hurting overall detail at 2khz

And at the end of it all, that's assuming your crossover was 500hz. That's really low -3db point for a 4" midrange driver, and that also means it's getting significant energy as low as 2-3 octaves below that. Think of how much 125hz energy it's getting. It might be 18db down on paper, but the power demands at these frequencies are equal to the power demands at 1khz in real source material.

And all of the above, is assuming these drivers are perfectly flat to even get these ideal slopes. How many 4" drivers are perfectly flat, without a vent, down to 65hz? Only ones with really low sensitivity....otherwise you won't be getting that "perfect phase" you wanted

 

[GranteedEV]

Moving away from center means more distance from one speaker, less distance to the other. Than means that you can easily go from 0/0 to +6db/-6db with just a few feet of movement. It doesn't matter what your speaker does, you can't change that.

I don't think you quite understand how cross-firing works.

The off-axis response of a narrow directivity speaker, falls off pretty quickly.

When cross-fired, you're getting more on-axis with the far side speaker (which compensates for distance losses) and more closer to the near-side speaker (which compensates for off-axis losses). I don't think this works for truly beaming speakers, but that's still different from a 90deg coverage pattern.

 

[cpp]

AES E-Library » On the Audibility of Midrange Phase Distortion in Audio Systems

if your a member the paper is $5.00

 

[randyb]

Upon closer inspection of some tests I've seen, they often seem more like a test of preference towards loudspeakers with more bass extension than overall fidelity. In a direct instantaneous comparison, I've personally found listeners gravitate towards the speaker with most boom and sizzle. This is why I maintain long term and frequent listening tests of products on an individual basis is necessary. If Speaker A has more bass than Speaker B, but Speaker B produces better mids/highs and can play at higher SPL more cleanly, than it may be a good idea to bass manage both speakers to a level matched sub and run separate listening tests to see how listener preference changes. Nobody does this to my knowledge.

.

That is one of the reasons I abandoned my blind test of bookshelf speakers. I thought the bass output would bias the participants and I realliy didn't want to test a bookshelf for bass output when a sub would be used (and mating a bookshelf to a sub is not that easy IMO and how do you do it fair for two different speakers that require different crossover points etc). That said you can test with a brick filter above a certain bass frequency. I think a lot of people would be surprised not at the difference but at the similarity.

I do agree with testing how loud they will play and the dynamics aspect. That is why a speaker like Procella or Geddes etc will walk all over the usual bookshelf without a compression tweeter. That is just my opinion and if you design a crossover with drivers to be more sensitive, aren't there tradeoffs?

 

[haraldo]

Dunlavy Closes Doors | Stereophile.com

And an interesting quote from one of the speaker reviews:

"DAL firmly believes that a full set of credible measurements, made by qualified engineering staff using state-of-the-art equipment and facilities, can reliably predict the potential of a loudspeaker to accurately reproduce the complex sounds of music."—Dunlavy Audio Labs

Dunlavy Audio Laboratories SC-IV loudspeaker | Stereophile.com

And a review that has off-axis measurements (that don't look great for that matter):

Dunlavy Audio Labs SC-IV/A loudspeaker | Stereophile.com

You should also read this arcticle, very interesting!
Loudspeaker designer John Dunlavy: By the Numbers... | Stereophile.com

 

[ski2xblack]

I do agree with testing how loud they will play and the dynamics aspect. That is why a speaker like Procella or Geddes etc will walk all over the usual bookshelf without a compression tweeter. That is just my opinion and if you design a crossover with drivers to be more sensitive, aren't there tradeoffs?

Tradeoffs? WAF. The waveguide is large, and the high sensitivity woofer is large, and the enclosure ends up being large to have any semblance of bass extension, so what you end up with has a very low WAF.

 

[GranteedEV]

Tradeoffs? WAF. The waveguide is large, and the high sensitivity woofer is large, and the enclosure ends up being large to have any semblance of bass extension, so what you end up with has a very low WAF.

The other one is that compression drivers aren't normally designed for hi fi, so even if they're still excellent, they weren't built from the ground up for the application.

 

[Nuance AH]

The other one is that compression drivers aren't normally designed for hi fi, so even if they're still excellent, they weren't built from the ground up for the application.

I agree completely. In my opinion you lose detail with the high efficiency drivers. The avid purist or 2-channel enthusiast isn't about to give that up. Finding a balance is key for HT and music lovers.

 

[GranteedEV]

I agree completely. In my opinion you lose detail with the high efficiency drivers.

Ehhh......That's just a broad generalization, subject to scrutiny. Is there even detail you lose from a Beyma TPL-150H compared to 99% of non-RAAL tweeters out there? How much detail do you lose from an Acoustic Elegance TD10M from 95% of midranges out there that aren't Seas Excels?

My only point is that you really need to make a compression driver / waveguide combination from the ground up. As long as you're using off-the shelf parts you're limited to them, and just every hi-fi company out there using these high efficiency drivers, is using off-the shelf parts. You're limited in quantity/selection for the application more than anything.

The choice ends up being TAD, or a cost-is-an-object compromise.

Now the problem with TAD, is that these drivers cost almost as much as a pair of Salk HT2-TLs. The reality is there's no waveguide speaker using TAD Drivers. Geddes did some testing of prototypes and people couldn't identify the TAD Summas from the B&C Summas reliably in a blind test.

Compression drivers have some serious advantages WRT hi-fi...

The main one, is that you can push the crossover rather low (IE 850hz and in the case of the TAD, ~600hz) with just a 1" tweeter. And of course there's the directivity control. These are the two main ones (with dynamics being tertiary but of course welcome)

There just isn't a speaker being sold that uses the best compression driver in an actual and optimal waveguide. It's funny, because that relates to the thread even though I went off on a tangent.

"Blind test with a sample of 27 people show the $150 mylar tweeter with 16khz breakup is equally preferred to the $2000 beryllium tweeter"

So do a 360 to the first page of this thread and you get to my issue with that

 

[haraldo]

If you've got access to it, the paper is called ""On the Audibility of midrange phase distortion .....(something)....". My university probably lets me access this stuff but i've never really tried. Here's a good substitute, though:

Human Hearing - Phase Distortion Audibility Part 2 — Reviews and News from Audioholics



Right, but I overall don't pay much attention to it. My point was, on the heirarchy of important stuff, where does step response sit?



As far as vertical off-axis response, any time you have a crossover(IE two drivers operating in the same frequency band with Center to Center spacing larger than 1/4 wavelength), you've sacrificed some. The more overlap in the crossover, the more sacrifice in vertical off-axis response. A 6db/octave crossover will have the least phase distortion, but the most overlap.

As far as horizontal off-axis response (and this applies to vertical as well), it depends on the width of a driver operating at a given frequency.

Even a 4" driver has narrowing dispersion at 2khz. So a shallow slope on it will make its narrowing dispersion play up higher. But also, to get a shallow slope on a tweeter, you'll need 2X, and more likely 3x the crossover point. So not only are you making the 4" driver play higher due to a shallow slope, you're rolling it off much higher. You're probably asking a 4" driver to play up to ~5-6khz as its -3db point, and much higher on a whole. And to add to that yet again, the higher crossover point sacrifices center to center spacing, which is another blow to the vertical off-axis response.

And then add that the 4" driver is playing much lower because it too, needs a shallow slope. Chances are in a 3-way (and more likely 4-way or even 5-way), there's a lot of bandpass gain. So actually, with shallow slopes necessary for phase coherence, you have a lot of drivers contributing in the same frequency region. So at 2khz, it's not just a 4 inch mid, and a 1" tweeter that's 8-10db down, but there's also a lot of bleeding in of the midbass driver, which you can repeat all of the above with respect to off-axis response. Let's say it's an 8" midrange driver crossed over at 500hz. That's also only around 12db down at 2khz. It's contributing, however slightly, to more on-axis response than off-axis response on a whole. Now add one more factor in. An 8" midbass driver that's only 12db down at 2khz, needs to not be breaking up. So you probably have to use a softer cone material that's probably hurting overall detail at 2khz

And at the end of it all, that's assuming your crossover was 500hz. That's really low -3db point for a 4" midrange driver, and that also means it's getting significant energy as low as 2-3 octaves below that. Think of how much 125hz energy it's getting. It might be 18db down on paper, but the power demands at these frequencies are equal to the power demands at 1khz in real source material.

And all of the above, is assuming these drivers are perfectly flat to even get these ideal slopes. How many 4" drivers are perfectly flat, without a vent, down to 65hz? Only ones with really low sensitivity....otherwise you won't be getting that "perfect phase" you wanted

Thank you for the incredible in-depth detail writing on this subject, it certainly provide some clue as to how difficult this is, and probably why many of these state-of-the art loudspeakers with first order crossovers are huge and with so many drivers... I never ever seen any proper DIY projects with first order crossovers, perhaps there's a good reason for this as being too bl¤¤#& difficult...

So utilizing DEQX or Behringer Ultradrive DCX2496? (read digital crossovers) that's phase perfect and higher order, will we not be able to get through these issues and get time/phase coherency without all the downsides?

Warranty of speakers will of course be thrown out the window.......

And as to these tests you refer to, they are mainly using Sawtooth Wave signals to test for audibility of phase distortion.... well, I never listen to Sawtooth Wave for entertainment or in any other way

 

[Nuance AH]

Ehhh......That's just a broad generalization, subject to scrutiny. I

Which I see you had no issues doing.

I stand by my statement but at the same time am in agreement with you: there's no affordable speaker that uses such drivers and still passes WAF (important for some people) that doesn't sacrifice detail and other various hi-fi attributes. For example, controlled directivity reeks havoc on the sound stage width and depth based on what I've heard. It simply blows my mind that the avid HT enthusiast claims speakers using these high sensitivity drivers don't give up detail, timbre accuracy or other hi-fi attributes. How they can't hear the same things me and my buddies do is beyond me, as it's so obvious to us. Bias is a strong deterrent, but to each their own, and YMMV, of course.

 

[randyb]

Which I see you had no issues doing.

I stand by my statement but at the same time am in agreement with you: there's no affordable speaker that uses such drivers and still passes WAF (important for some people) that doesn't sacrifice detail and other various hi-fi attributes. For example, controlled directivity reeks havoc on the sound stage width and depth based on what I've heard. It simply blows my mind that the avid HT enthusiast claims speakers using these high sensitivity drivers don't give up detail, timbre accuracy or other hi-fi attributes. How they can't hear the same things me and my buddies do is beyond me, as it's so obvious to us. Bias is a strong deterrent, but to each their own, and YMMV, of course.

It will be interesting at the WI gt to see how the speakers represented do when driven hard. So far with few exceptions, I find that as loudness increases the music starts to sound compressed. Imaging is a different matter, and will depend on the design etc. My use of compression speakers (Procella's) is for multi channel music and movies, hence why I ordered the Phil 2's. My audition at Dennis's was they sounded amazing but at really loud levels, I thought I heard the start of some compression of peaks. That being said, it could also have been inherent in the recording so I will reserve judgement. It certainly did not disuade me from wanting them for 2 channel music.

I respect those that say a "good speaker" is good "good speaker" for both music and movies. However, I have to respectfully disagree. I think when you are watching a movie you are aware of the sound but you are not as critical (sort of a texting and driving thing) except when things get loud and if the sound compresses the dynamic range (or lack thereof) it is noticeable. Now, I am talking about listening at fairly loud levels. The thread on AVS that talks about SPL's and list high SPL speakers has a really nice post on about the last or second to last page that I agree with. And remember, imaging with surround sound bases a lot on the center and surround speakers and a "good imaging speaker" for HT really doesn't make sense to me.

I certainly agree that bias is inherent in all we do but that works two ways.