Crossover Phase Behaviour

[Bill13]

this thread split off from here Audience A3



I need some 'expert' help regarding idea of using RAAL ribbons (with Audience A3s as midrange array, or maybe cheaper Fountek 3" as midranges). Of course this assumed one would even want to try adding a tweeter to full-range drivers that already offer good trebles.
Maybe I should post this elsewhere on this site -- where?

IF it's possible, how can a high-order (e.g. 4th order) passive speaker crossover be designed that maintains flat amplitude response AND, at the same time, maintain linear-phase coherency through the crossover region, assuming reasonably wide-bandwidth drivers are employed that have extended response above & below the 4th order crossover frequency ---

I was astounded to read YG Acoustics' advertisement in the current issue of TAS. Also, there's a review of the Kipod Studio Loudspeaker (TAS 199). YG Acoustics designer claimed that he developed new software that simultaneously optimizes 4th-order crossover's amplitude and phase such that deviation from perfect phase linearity is only 5 degrees -- and that the total phase rotation for the 4th order crossover slopes from 200 Hz on up is only 360 degrees! Heretofore, impression was that such linear phase reponse was essentially "impossible" to achieve with conventional higher-order passive crossovers (like L-R 4th order). Theoretically can get coherent phase shift with first-order crossover approach, but I don't see how YG Acoustics can get 4th order rolloff crossover with a single 360 degree phase rotation for the speaker response from 200 Hz up.
Doesn't seem to obey electrical network theory.

Puzzle: Maybe somebody can explain how YG Acoustics combines 4th order crossover slopes while at the same time gets only first-order phase rotation?
Guess that's why they say that their YG speakers are the 'best in the world'?

[chrisb]

Originally Posted by Bill13

I was astounded to read YG Acoustics' advertisement in the current issue of TAS. Also, there's a review of the Kipod Studio Loudspeaker (TAS 199). YG Acoustics designer claimed that he developed new software that simultaneously optimizes 4th-order crossover's amplitude and phase such that deviation from perfect phase linearity is only 5 degrees -- and that the total phase rotation for the 4th order crossover slopes from 200 Hz on up is only 360 degrees! Heretofore, impression was that such linear phase reponse was essentially "impossible" to achieve with conventional higher-order passive crossovers (like L-R 4th order). Theoretically can get coherent phase shift with first-order crossover approach, but I don't see how YG Acoustics can get 4th order rolloff crossover with a single 360 degree phase rotation for the speaker response from 200 Hz up.
Doesn't seem to obey electrical network theory.

Puzzle: Maybe somebody can explain how YG Acoustics combines 4th order crossover slopes while at the same time gets only first-order phase rotation?
Guess that's why they say that their YG speakers are the 'best in the world'?

If it's in an advert, it must be true

the best in the world?

really?


never heard that before

what about the inevitable upgrades?

all cynicism aside, they might well sound very nice indeed, but am I the only one long since bored with this type of hyperbole?

[planet10]

Doesn't a 4th order LR have only 360 degrees of phase rotation?

There are ways to maintain continuous phase response using high-order active analog filters (LR AES paper use of all-pass filters), but reports are that the cure is worse than the disease.

With small FRs with very good HF response, and a tweeter with good LF response, one should have the overlap needed (if not the ability to get the drivers close enuff together) to use a 1st order XO.

dave

[Bill13]

Dave,

I believe you are correct about 4th order LR having 360 degrees rotation through crossover region (typ. 90 degrees shift per network pole). However, as I recall, YG Acoustics appears to be unique in maintaining only 5 degree error from linear phase characteristic for their symmetrical-slope passive crossover network = constant group delay in a fourth-order LR-like crossover, even when including all the crossover network frequency response (driver) compensation adjustments (this is from my poor memory of the applicable crossover theory - which admittedly is rusty -- the reason why I would appreciate other viewpoints). Phase rotation over the entire 200 Hz on up speaker bandwidth seems to be more comparible to first-order speaker designs (e.g. Thiel) -- again please see the TAS YG Acoustics advertisement.

Vance Dickason's "Loudspeaker Design Cookbook" does present formulas for a so-called "forth-order linear-phase" passive crossover. He says that the asymetrical derivation makes this particular filter's suitability questionable.
This crossover's name sounds very intriguing, but still, the summed group delay for this forth-order linear-phase" crossover is not equal or better than a first-order crossover (I understand).

In the TAS advertisement the YG Acoustics claim of only 5 degree phase nonlinearity and perfomance superior to a first-order crossover is remarkable but very hard to believe (for me, anyway).

-- wish I could duplicate the YG Acoustics 24 db/oct with constant group delay approach for a RAAL ribbon tweeter!

Such a small phase linearity error means that all frequency components are 'coherently' radiated all at the same time (constant group delay). Why can't the other high-end speaker companies achieve these coherent 4th order slopes too?

reference: Thiel Inc. first-order crossover systems are designed to provide phase coherent transitions between very wide bandwidth drivers. Thiel seems to feel that only first order passive crossovers can do the job.

[planet10]

Originally Posted by Bill13

In the TAS advertisement the YG Acoustics claim of only 5 degree phase nonlinearity and perfomance superior to a first-order crossover is remarkable but very hard to believe (for me, anyway).

Me too. Keep in mind that there are a number of phase related terms that have different meaning.

I have read everything i can about linear phase XOs, and it seems to me that you are still stuck with 1st order active or passive or subtractive active (which have their own set of issues).

My solution to the problem is to have either no XO or one below 400 Hz.

dave

[Bill13]

Dave,

Thanks for your reply.

Sorry, I didn't understand what you said in your post: " -- or passive of subtractive active (which have their own set of issues)" -- could you clarify this -- it would help me to understand more about linear phase XOs.

Anyway, I gather that you don't believe claim that YG Acoustics can achieve constant group delay, while at the same time, achieving symmetrical 24 dB/oct crossover slopes ?

If they really can do this, it would be a wonderful idea for us DIY speaker hobbyists to emulate.

[planet10]

it should read passive OR subtractive... i never did learn to type.

dave

[planet10]

Originally Posted by Bill13

Anyway, I gather that you don't believe claim that YG Acoustics can achieve constant group delay, while at the same time, achieving symmetrical 24 dB/oct crossover slopes ?

I believe that they cannot achieve a continuous function with 4th order XOs.

dave

[ScottG]

Originally Posted by Bill13

In the TAS advertisement the YG Acoustics claim of only 5 degree phase nonlinearity and perfomance superior to a first-order crossover is remarkable but very hard to believe (for me, anyway).

Here is the Stereophile add. It isn't a minimum phase design.

http://www.ygacoustics.com/YG_Acoust...e_May_2008.pdf

They talk about the lowest "relative" phase.

Stereophile: YG Acoustics Anat Reference II Professional loudspeaker

Looking at the step response shows what's going on.

as for "the best loudspeaker on Earth"..

Looking at the Burmeister B99's linearity seems to suggest otherwise:

http://stereophile.com/floorloudspea...01/index6.html

[Bill13]

Thanks very much for your insightful observation regarding the Stereophile review's measurements: "Fig.8 YGA Anat Reference II Main Module, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth)".

This initial rise of the step response waveform 'could have/should have' been improved by time-aligning the acoustic-centers of the drivers?
Too bad, since the use of a YG Acoustic speaker's tweeter-waveguide may have been used much more advantagely to offset the tweeter behind the woofer - so I think, so far, at first impression ...

Thanks again,

Bill