[csero]

Davey,
     An interesting note was that Jack from NHT said that time and overall phase is so subjectively hard to pick up that the tweeter/midrange was time aligned (from memory) "just because there was no reason not to do it".  Kevin Voecks and many other excellent speaker designers have simply been unable to verify that phase/time is very audible, if at all, and Voecks has done some serious looking into it.  But, like I said, digital makes it easy, so why not?  Driver/cone distortion, dispersion and oth ...

I can only repeat myself: the Blauert and Laws criteria
For audibility the thresholds are approximately the following:

Frequency Group delay
8 kHz          2 msec
4 kHz          1.5 msec
2 kHz          1 msec
1 kHz          2 msec
500 Hz        3.2 msec
below that it is even higher

Note: 1 ms is about 1 feet difference in the distance of the acoustic centers of the drivers

[John Casler]

Scotty/Davey,

Thanks for confirming.  I was pretty sure it was a single point.  That is why I never understood sloped baffles.  Simple geometry would tell you that that will only work for a single point, which is all well and good, but a different claim then often made by the manufacturers.

Truth be told, if one is evaluating critically a speaker design then they should be in the listening sweet spot and not walking about the room.  But if the speaker does its thing for a spot 8.8' away from the speakers and another listener happens to have his listening spot 9.2' away, what then?  I guess that is why digital correction is such a cool thing, because it allows flexibility in the design to match the room and setup of the user, provided the user knows what they are doing. ...

Hi Josh,

The sloped baffle to my understanding has its greatest precision needed to listening "height".  Being the exact height provides the angular alignment for the slope to be proper.

This means that the there are now "two" EXACT positional requirments:

1) Equidistance from each speaker (horizontal sweet spot)
2) Exact angular incidence "height" (vertical sweet spot)

And while "digital correction" can possibly allow you to adjust this for your actual position (rather than adjusting your position, you adjust the signal) it still requires that onece set, that you are in the exact vertical and horizontal positions or it doesn't work.

Or at least that is how I perceive it.

[LordCloud]

I am in no way able to comment technically on crossover design. I can only tell you that out of all of the speakers I've ever heard, only ones with 1st order crossovers, wired in phased, with sloped baffles, give the actual space, image density, and soundstaging that are essential parts of the puzzle for me. It just doesn't make much sense to me to have multiple drivers starting and stopping at different times, on purpose. I know that no speaker design is perfect, but time and phase coherent speakers just sound more real to me. Just my 2 cents.

[ctviggen]

After listening to many speakers, I chose two (Linn and VMPS), and neither of them have sloped baffles.  I don't think a sloped baffle means much if anything.  I've also not seen a good definition of time and phase coherence, other than square waves.  I have to say that I'm personally not impressed by Dunlavys, but many people like them.  (Which are supposedly great at producing square waves but don't have sloped baffles.)

[LordCloud]

Dunlavys don't have sloped baffles but their drivers are staggered. The baffles are sloped or the drivers staggered so that the drivers are in the same plane acoustically. With many speakers because of the high order crossover and the non sloped baffle the sound from the tweeter leaves the speaker before the other drivers and therefore makes a speaker seem more detailed, and most people like that. I believe it is the frequencies leaving the speaker baffle at the same time that give time and phase coherent speakers their unmatched(in my experience) soundstaging, imaging, image density, and detail retrieval. Again, I have limited knowledge of the technical side of things, but I know what I hear.

[John Ashman]

With many speakers because of the high order crossover and the non sloped baffle the sound from the tweeter leaves the speaker before the other drivers and therefore makes a speaker seem more detailed, and most people like that. .

No cause/effect relationship to that at all!

[AndrewH]

Wow this topic took off again.

John Ashman - fancy running into you here  Why do you say there is no cause/effect relationship between hearing disjointed treble to the 360 degrees of phase rotation in a 4th order crossover?

As John Kreskovsky describes on his website:


These  figures corrispond to the 4th order Linkwitz Riley crossover with both drivers connected with positive polarity. As with the 3rd order Butterworth case  the impulse and step response are clearly distotred. The step response again shows that there is sufficient delay introduced by the crossover so that the response of each driver is individually apparent. The  little difference between the flat baffle and the case where the driver ACs are aligned again indicates that the crossover introduces more delay than  that represented by the 30mm relative offset in the missaligned case  

http://www.geocities.com/kreskovs/TimeAligned2.html

[LordCloud]

One thing I've noticed in my times as an audiophile is that too many of us love sound and not the sound of music. Many audiophiles have no idea what music is supposed to really sound like, but they can make a system sound really exciting. "Listen to that bass", or " wow that tweeter is really good". That's because these frequency ranges are thrown at you seperately and so you hear them as such and not as a musical event. I don't know how it's possible to piece together the illusion of actual performers in an actual space if the pieces to that puzzle all come at you at different times.

[AndrewH]

LordCloud - I have to agree with you.  But it's also true that each audiophile thinks his hearing is superior to the other's. Myself, I have plenty of experience with playing and listening to live musical instruments.  I think *playing* the instrument helps give a better sense of what to listen for when that instrument is being performed by someone else, such that one is able to discern when the instrument doesn't sound quite right.  That's pure opinion though, might be good for an editorial at PhaseCoherent.com.  Actually such an experiment could be pretty easy to set up.  If I had more time and funds I could pay participants and publish the results.

[LordCloud]

You know, it's not so much that my hearing is better than anyone else's, I would say that there are plenty of guys that hear far better than I. I think it's that many audiophiles don't know what to listen for. I'm willing to bet that the reference for most ausiophiles is other audio systems. And those other audio systems just pick apart frequencies as opposed to bringing music to the listener. Of course everything I say is only opinion as well. I really don't want anyone to get the idea that I think I'm some audio guru or anything like that,  I'm just relaying my experiences.

[John Ashman]

Wow this topic took off again.

John Ashman - fancy running into you here  Why do you say there is no cause/effect relationship between hearing disjointed treble to the 360 degrees of phase rotation in a 4th order crossover?

Uh, because we don't listen to impulse signals.  If you hear disjointed treble, it's because of a bad choice of drivers and poor crossover design.  I've heard "phase correct" speakers with disjointed treble and non-phase correct ones that don't.

[LordCloud]

By design, isn't a non time and phase coreect speaker's  treble disjointed? As well as the bass and midrange.?

[Davey]

"Disjointed" being a subjective term?
I find myself agreeing with John on this one.  If speakers exhibit this "quality" it's most likely related to some other aspect of the design rather than "time/phase coherence."  Objective measurements can show the situation readily, but the real question is about our perception and how we "process" waveform distortion.

The arguement/discussion of audibility of phase distortion goes back decades and there are many excellent papers on the subject....We're certainly not going to come to a definitive conclusion in this ridiculously titled little thread on AC in 2005.  

Generally, folks who tout the excellent qualities of certain "phase coherent" speakers are usually basing this preference on inadequate information and/or apples/oranges listening/testing methodology.  Say a person really likes the sound of Dunlavy or Thiel or Meadowlark speaker....aha it must be because of the first-order crossovers and phase-coherent design right?  No, it's not nearly that simple.  There are many more factors involved.  The only way to come to even a somewhat half-baked conclusion would be to redesign each of those speakers with alternative crossovers that were not linear-phase and judge the results with a blind test.
In fact, I think even the speakers need to be taken out of the loop when making determinations about phase distortion.  Headphones are a much better test instrument with the phase distortion then added in at line-level via a hardware circuit or added to a re-mastered CD with modified and original tracks.  Only then can you really zero in on the audible effects of JUST phase-coherence.

The only valid testing scenario I've seen for this was put forward by Linkwitz awhile back.  It isolates the audible effects of phase distortion as well as any setup I've come across.  Unfortunately few/no people were interested or even understood the concept.  Oh well.

http://www.linkwitzlab.com/phs-dist.htm

I think John's main point is that regardless of the audibility of this phenomenon the DSP tools now exist to "correct" it without too much trouble.....so why not do it.  You can't argue with logic like that.  

Cheers,

Davey.

[John Ashman]

By design, isn't a non time and phase coreect speaker's  treble disjointed? As well as the bass and midrange.?

One would think (listen to Davey, he is very wise ).  However, while having proper time alignment and phase is a plus, it's really the driver/cabinet/crossover design and a 1st order design can have even more problems with integration at times.  You see, if a midrange driver has sufficiently wide dispersion, is sufficiently accurate and sufficiently low distortion at the upper end of its range, there is no audible crossover sound, even if its not phase correct or time aligned.  You might be able to pick up some slight acoustic distortion or roughness, but it would be over such a wide area that it couldn't possibly be picked up subjectively as being disjointed and this type of disortion is surprisingly well surpressed by our ear/brain, far less apparent audibly than it is on a graph.  Vandersteen 3A tweeters sound disjointed to me.  How would you explain this?  Although you could make a technical argument that a driver must be phase/time aligned to be "jointed", we just don't hear it that way.  Usually it's a dispersion problem, distortion problem and/or a FR problem.  

I am not arguing *against* time alignment.  If a digital speaker weren't time/phase aligned, I'd be complaining about it - there's simply no downside or cost to do this in the digital domain.  But with a 6dB/octave crossover, there are *multiple* downsides and if you can't hear them, buy such a design.  I *can* hear it and every single 6dB/octave speaker I've heard looks far better on paper than it sounds and they tend to squeal like a pig when you turn them up.  I attended a Thiel demo that was *horrible* because the tweeters (and metal midranges) were really distressing playing HT material.  I had to leave.  Vandersteens go from warm and polite to shrill when you turn them up.  I had to sell a pair of Vandersteen 3As for $500 because *no one* would buy them because a certain $1000/pr retail speaker just plain sounded better to everyone that heard them (I sold a lot of $1000/pr speakers though!).    

And then, of course, it costs more money to make a time aligned cabinet.  I personally think Vandersteen is completely full of crap when they say their cabinet is so little to build.  There's a *lot* more time and cost to build that than a simple V-cut/Fold box.  Thiels are clearly expensive.  I'm sure Meadowlarks are as well.  But that impacts the performance/value.  If you double the price to make a fancy cabinet, does the performance go up commensurately? Not usually.   And, while a 1st order crossover *should* be a *very* simple design, check them out, they all have a ton of driver compensation in there.  LOTS of stuff for the signal to pass through.  

So, first order, time aligned, phase correct speakers have the following problems:

1.  Increased cabinet costs
2.  Extremely poor vertical dispersion
3.  High cone woofer/midrange cone resonances
4.  Increased driver distortion for midrange/tweeter
5.  Limited dynamic range
6.  Very limited sweetspot, vertically and horizontally
7.  Increased crossover complexity and expense.  
8.  Limited choice of appropriate drivers for good performance

All to get ONE feature that is highly debated at best and subjectively not terribly significant unless you're predisposed to believe that this is the end all/be all of speaker design.  If time/phase alignment comes with out cost, that is one thing.  But to trade many areas of performance for an improvement in one makes no real sense.  But it does add to the wide pallette of speakers available, I just don't think I'd dedicate a website to it because you're just asking for abuse

[Dan Banquer]

Maybe some of the speaker experts can help me here: Why is it that linearity is never discussed when it comes to loudspeakers? I hear a lot of things about imaging, staging, and acoustics but I don't recall people talking about linearity when it comes to loudspeakers.
Having some background in electronics and in audio I would think impulse response would be able to tell us something about the linearity of the loudspeaker. Would it not?
           d.b.

[John Ashman]

People don't talk about it because it's hard to address linearity.  You can choose a driver for its linearity.  But speakers are terribly non-linear creatures.  Motor distortion, phase/time, compression, etc cause issues here.  BUT, just like servo can make a driver operate in a more linear fashion, DSP will soon start to address non-linearity issues in the near future, something that I have heard is being worked on now.

[Dan Banquer]

"People don't talk about it because it's hard to address linearity"
Exactly!
I think DSP will be of some help, but for those who get involved in engineering one generally finds that solving the problem at it's root has far better results than attempting to fix the mistake later on down the line.
This maybe a rather bold statement, but I think if loudspeakers were more linear devices than a whole lot of DSP would not be necessary.
                  d.b.

[John Ashman]

DSP does help speakers already be more linear by removing strain off the drivers, doing time/phase alignment and a host of other things so it *indirectly* helps the drivers be more linear and also often allows a designer to choose a driver that may be more linear, but have other issues solvable with DSP.  But they say they can do more advanced processing to correct for driver non-linearities and that is good.  I think servo could also creep back in and into higher and higher frequencies.  

I want to see *laser* servos!

[Dan Banquer]

I have my doubts about feedback and servos in this application. The premise of this relys upon the the driver mechanism to be fast enough to respond. Is that an assumption one can make using a loudspeaker driver?
          d.b.
BTW: I note no response to my previous question on impulse response. Am I missing something?

[John Ashman]

Well, all I know is that DEQX says that they correct for the impulse response of the speaker. We'll probably have a better picture of what it can do when NHT's Xd gets put under a microscope.