Time/Phase Alignment, Acoustic Center, Lobing etc. - new tech note from Purifi

[boXem]

Still have to digest it before commenting

https://purifi-audio.com/blog/tech-notes-1/time-phase-alignment-acoustic-center-lobing-etc-14

 

[Salt]

Stumbled over this a few days ago and wondered while reading when the breaking news may occur ... didn't find it actually.

 

[Pearljam5000]

How good are Wilson speakers in that regard ?

 

[pma]

Still have to digest it before commenting

https://purifi-audio.com/blog/tech-notes-1/time-phase-alignment-acoustic-center-lobing-etc-14

Thanks for the link . That’s a good one, and interesting coincidence.

 

[Curvature]

How good are Wilson speakers in that regard ?

They make huge sacrifices in directivity for meaningless gains in step response, which is moreonly only good for one listening position.

 

[pma]

Some examples of vertical "lobing" at crossover freq with ideal point sources and various filter types and distance of sources.

 

[Salt]

Still have to digest it before commenting

https://purifi-audio.com/blog/tech-notes-1/time-phase-alignment-acoustic-center-lobing-etc-14

Medium intake to gut outcome passage is 2 and a half hour.
So...?

 

[thewas]

They make huge sacrifices in directivity for meaningless gains in step response, which is moreonly only good for one listening position.

And even then not for the reflected off-axis sound.

 

[FrantzM]

How good are Wilson speakers in that regard ?

From the posted article by Purifi.-Audio ....

What is a so-called time-aligned speaker?​

Even if you use method 3. Where the baffle is designed to compensate for the latency time difference of the drivers we will, contrary to our intuition, not generally get a linear phase (aka pure delay) where all frequencies arrive at the same time. The reason is that the sum of the low and high pass filters of the textbook crossovers result in an all-pass response with a generally nonlinear phase. Using analog filters, only a first order crossover sums up with linear phase. In addition, there are a few exotic analog solutions that result in linear phase: one being the filler driver by B&O from the 1970’ies which uses an extra driver driven by a band-pass filter (their top model used 5 drivers). Another exotic solution is the class of delay-derived filters by Lipshitz and Vanderkooy. However, they require a pure delay which is extraordinarily difficult to do in analog. In practice, true phase linear speakers using higher than first order crossovers can only be done using complex DSP filtering (e.g., Kii Three) .

Consequently, it is a misunderstanding that method 3 (equalization of the latency times by physical design of the baffle/waveguides) results in a so-called time-aligned loudspeaker in the sense of being linear phase.

Peace.

 

[gnarly]

From the posted article by Purifi.-Audio ....

What is a so-called time-aligned speaker?​

Even if you use method 3. Where the baffle is designed to compensate for the latency time difference of the drivers we will, contrary to our intuition, not generally get a linear phase (aka pure delay) where all frequencies arrive at the same time. The reason is that the sum of the low and high pass filters of the textbook crossovers result in an all-pass response with a generally nonlinear phase. Using analog filters, only a first order crossover sums up with linear phase. In addition, there are a few exotic analog solutions that result in linear phase: one being the filler driver by B&O from the 1970’ies which uses an extra driver driven by a band-pass filter (their top model used 5 drivers). Another exotic solution is the class of delay-derived filters by Lipshitz and Vanderkooy. However, they require a pure delay which is extraordinarily difficult to do in analog. In practice, true phase linear speakers using higher than first order crossovers can only be done using complex DSP filtering (e.g., Kii Three) .

Consequently, it is a misunderstanding that method 3 (equalization of the latency times by physical design of the baffle/waveguides) results in a so-called time-aligned loudspeaker in the sense of being linear phase.

Peace.

Yes, it was nice to see an article that clearly discussed the difference between pure time delay, aka time alignment, and phase.

I wonder however, how many people associate the term "time-aligned" with linear phase.
Maybe more than i think, because I also wonder how many people fully realize what linear phase means.....
and that the inevitable phase rotation of traditional IIR crossovers prohibits linear phase. (IIR xovers beyond first order)

Anyway, I do wish the excellent article went on to talk about how simple it is to use DSP to get pure time delay correct down, to the precision of the time of a single sample.
or how simple it is to implement complementary linear phase crossovers (of any order).

With DSP fixed time becomes alignment become much easier; phase alignment absent rotation becomes much easier.
Nothing really complex about it.
If anything, using traditional IIR is far more complex..... ime.

So I think to say 'true linear-phase speakers using higher than first order crossovers can only be done using complex DSP filtering' is a gross overstatement.

Maybe the author meant the correction of the all-pass nature of the IIR crossovers with DSP is complex....and with that I can agree.
But if that's the case, I ask........ if the DSP capable of that type of phase correction, which means FIR, why not just use linear-phase crossovers to begin with? (and ditch all the complexity)

 

[boXem]

Yes, it was nice to see an article that clearly discussed the difference between pure time delay, aka time alignment, and phase.

I wonder however, how many people associate the term "time-aligned" with linear phase.
Maybe more than i think, because I also wonder how many people fully realize what linear phase means.....
and that the inevitable phase rotation of traditional IIR crossovers prohibits linear phase. (IIR xovers beyond first order)

Anyway, I do wish the excellent article went on to talk about how simple it is to use DSP to get pure time delay correct down, to the precision of the time of a single sample.
or how simple it is to implement complementary linear phase crossovers (of any order).

With DSP fixed time becomes alignment become much easier; phase alignment absent rotation becomes much easier.
Nothing really complex about it.
If anything, using traditional IIR is far more complex..... ime.

So I think to say 'true linear-phase speakers using higher than first order crossovers can only be done using complex DSP filtering' is a gross overstatement.

Maybe the author meant the correction of the all-pass nature of the IIR crossovers with DSP is complex....and with that I can agree.
But if that's the case, I ask........ if the DSP capable of that type of phase correction, which means FIR, why not just use linear-phase crossovers to begin with? (and ditch all the complexity)

Although the article was written by Lars Risbo, "Mr. Filter" at Purifi is Bruno Putzeys and I can't believe that he didn't participate. The allusion to the Kii 3 didn't come by chance. He explains his stance against FIR filters in the paper written for the Grimm LS1. https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf. If I understand well, the issue would be the introduction of pre-ringing and pre-echo.

 

[Salt]

Therebid nothing new about correct time and phase.
In the 70th of last century Joseph Manger invented the MSW for that reason, and that's almost 50 Years past.
Additionally, with speakers, the environment may take the greater part of performance than pre- or postringing or absolute phase alignment.

 

[boXem]

Therebid nothing new about correct time and phase.
In the 70th of last century Joseph Manger invented the MSW for that reason, and that's almost 50 Years past.
Additionally, with speakers, the environment may take the greater part of performance than pre- or postringing or absolute phase alignment.

Since when a tech note is supposed to bring advances to the science?
Pre-post ringing/echo is actually audible for speakers.

 

[fpitas]

Since when a tech note is supposed to bring advances to the science?
Pre-post ringing/echo is actually audible for speakers.

It can be. And in my experience, phase-flattening had no audible effect on an LR4 crossover at 800Hz. I can see that phase flattening might become justifiable if very high order crossovers are desired.

 

[gnarly]

Although the article was written by Lars Risbo, "Mr. Filter" at Purifi is Bruno Putzeys and I can't believe that he didn't participate. The allusion to the Kii 3 didn't come by chance. He explains his stance against FIR filters in the paper written for the Grimm LS1. https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf. If I understand well, the issue would be the introduction of pre-ringing and pre-echo.

Yes, I have to believe Bruno had a hand in the paper too. I've been a "student" of Bruno's, following his posts back from the days on Prosoundweb.
And am familiar with the Grimm LS1 paper.

The paper, imho, is an example of of what I see as needless complexity, really.
The part of it that describes the 'train wreck', the case against FIR, appears to be using simple impulse inversions of single on-axis responses.
Which is pretty much guaranteed to cause grief off-oxis, ime.
The case to avoid FIR extrapolates too far.....and a bit unfairly i think.

Then, the part that talks about the 'icing on the cake' later in the paper, makes a case for using FIR to correct the phase rotation of the IIR crossover in place.
This is what I was saying in my previous post saying makes no sense to me......if one has the FIR capability to make that LR4 IIR all-pass phase correction the paper describes, it means the FIR capability is there to just put in a linear-phase LR4 crossover to begin with, and skip the complexity.

 

[boXem]

Yes, I have to believe Bruno had a hand in the paper too. I've been a "student" of Bruno's, following his posts back from the days on Prosoundweb.
And am familiar with the Grimm LS1 paper.

The paper, imho, is an example of of what I see as needless complexity, really.
The part of it that describes the 'train wreck', the case against FIR, appears to be using simple impulse inversions of single on-axis responses.
Which is pretty much guaranteed to cause grief off-oxis, ime.
The case to avoid FIR extrapolates too far.....and a bit unfairly i think.

Then, the part that talks about the 'icing on the cake' later in the paper, makes a case for using FIR to correct the phase rotation of the IIR crossover in place.
This is what I was saying in my previous post saying makes no sense to me......if one has the FIR capability to make that LR4 IIR all-pass phase correction the paper describes, it means the FIR capability is there to just put in a linear-phase LR4 crossover to begin with, and skip the complexity.

My understanding of FIR filtering is that it introduces a delay. When the application is different from home music reproduction, delays can be problematic. The advantage of the IIR filters + global FIR would then be that the FIR can be disabled when required by the application.
I have to admit that I am not Mr. Filter myself so apologies if I am wrong.

 

[voodooless]

So, who dares to show us microphone recoding of a 1 kHz square wave over his speakers?

 

[Salt]

Since when a tech note is supposed to bring advances to the science?
Pre-post ringing/echo is actually audible for speakers.

Did not mention advantance to science (but cold coffee), neither neglect of ringing audible with speakers, but that acoustics of the listening room my be of higher influence to the listening experience.
Btw here we are the output; someone willing to discuss once more ringing at input (DAC and slt)?

 

[pma]

So, who dares to show us microphone recoding of a 1 kHz square wave over his speakers?

I am afraid of mostly misleading conclusions from such record, here .

 

[fpitas]

I am afraid of mostly misleading conclusions from such record, here .

Well that, and you'll only get a square wave in one tiny spot.