|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
24.13.228.29
In Reply to: RE: Absolute phase/polarity posted by Inmate51 on August 03, 2013 at 15:38:46
You are correct, the late Richard Heyser (passed away about 27 yeas ago) both proposed a way to measure acoustic phase but also related that to how phase shift alters a loudspeakers behavior in time.
You can probably find some of his papers on line like “determination of loudspeaker arrival time” Audiophiles debate this topic because how loudspeakers work can be confusing if you don’t have a way to measure what they do.
Magazines generally are not sources of technical information, they are marketing agents. The phase shift is not a feature, it is more or less unavoidable without dsp and so isn’t discussed very much.
We are not that sensitive to absolute phase in part because loudspeakers themselves in general do not preserve time or phase very well.
For example, a vented box exhibits a 360 degree phase rotation in the span from above to below the low frequency corner.
This means the lower frequencies are always delayed, as much as one entire cycle going from well above to below the lf corner in it’s response. This is why a sealed box preserves time better than a vented box with the same lf corner, the sealed box has half the phase shift around its low corner
The phase shift alters the harmonic time structure of a complex signal, a signal fed into it appears to be produced at different times which is not what the input signal did.
Making it more complicated, a perfectly located upper and lower drivers have a crossover to separate the frequencies. Above a first order filter ALL the normal crossovers types like Butterworth, Linkwitz and so on ALL exhibit phase shift going from above to below the crossover point.
For example, a 4th order filter also has 360 degrees of phase rotation and so the lower frequency range emerges later in time than the hf portion. It is not uncommon for a traditional loudspeaker to have hundreds even thousands of degrees of acoustic phase shift top to bottom and so the idea of preserving the input signal is out of the question.
If you can't produce all the frequencies in question at the same instant in time, then you have no way to judge absolute phase because the loudspeaker can’t preserve that over a significant bandwidth.
In spite of other shortcomings, this is why small full range speakers are popular, they don’t normally show that degree of crossover phase shift and so do a better job preserving the time related things like acoustic phase.
A loudspeaker that has flat amplitude and phase over a broad band, is the only type that can actually reproduce the wave shape of the input signal and then a square wave is a wave shape that is easily recognizable on an oscilloscope and in electronics was often used as a test signal..
VERY VERY few loudspeakers preserve time well enough to reproduce a square wave over a broad bandwidth and over a range of radiation angles.
Hope that helps
Tom Danley
Follow Ups:
~!
The Mind has No Firewall~ U.S. Army War College.
...a book that's been on my audio shelf for twenty-five years: The Wood Effect. Imagine, twenty-five years ahead of its time! There's a section on Richard Heyser in it too, including an interview IIRC. Also it chases down nearly every reference to absolute polarity in print until then, starting with an article JASA and even including double-blind tests! (For all the good that ever does... lol)
P.A.
PS It ain't cheap below, so leave me a message here for possible word on a hidden stash.
There is a test for transducers. Loo at Stereophile's Step impulse test and measurements. Stereophile, themselves downplay the results, uttering such platitudes like the "downward slope of the tweeter blends smoothly into the inverted midrange driver".
Such nonsense. In their own (Atkinson's) explanation, the graphed measurement should look basically like a right triangle: with a sharp rise time and a slow decay. Aktinson, himself, writes That he has seen such graphs only on ten (!) models from 5 manufacturers (Thiel, Vandersteen, Quad, Dunlavy, and Spica.
With even well regarded systems like Wilson displaying such appalling performance, it is no wonder that absolute polarity can not be heard by most, Inverting polarity basically become half a dozen of one and six of another.
One notable speaker designer put it very well ( he is on the Atkinson list, BTW). And I quote "if perfect frequency response and time alignment and phase alignment were the goal of every manufacturers and designer, shouldn't there be a convergence of sound at a particular price point?"
I am fortunate that haved owned Spicas, Vandersteens, Quads, and Thiels. For me absolute polarity is clearly audible, because I modify my speakers within 90 degrees of being time aligned.
Of course YMMV
Yup, lots of insight and good info you wrote there, Tomservo.
With regard to multi-way speaker systems, I'm hoping - as I finalize the design of my new ones - that I'll be able to ($$) incorporate appropriate time delay with the crossover in order to reduce some of that phase delay problem (they'll be tri-amped). I know Rane and dbx have some capabilities in that area, but I haven't looked into it in a couple years - since there have been other issues to work out, and a more-or-less normal life to lead.
Heyser really knew what he was doing. It's so great when a person almost intuitively understands the acoustical, mechanical and electrical physics of audio, AND is in a position to produce and publish high quality research. He and a couple handfuls of other folks were the giants who got us to where we understand most of what we do in audio.
Can y'all say TDS and TEF?
:)
You can't fix a group-delay problem with a single time-delay for a given band. The delay is frequency dependent. Some DSP/active boxes have phase correction, and the Thuneau Allocator computer-based DSP crossover has phase correction which it does with what they call "forward/reverse processing". It's kind of cool. By virtue of a latency delay, the process effectively "goes forward in time" (a trick using the latency delay) and causes in effect the opposite of group-delay. Sort of a group-"opposite of delay". Then, the crossover function causes the expected group-delay which in effect cancels out. So yes, you can make pretty square waves with standard filters such as 3rd order Butterworth and 4th order Linkwitz riley. The only trick is you need to know the acoustic function and Fc points for all crossover points, and the -6db point for the LF band, if ported. (This is to do with the group-delay of the ported design Tom was referring to).
Fascinating stuff.
Cheers,
Presto
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: