How far back?

[Edgar]

9 minutes ago, Chief bonehead said:

Oh I see. Different animals. I only deal with getting two or more sections to get them not only time aligned but “period” aligned to that the amplitudes at crossover add up to the proverbial 6 db while eliminating as much interference at crossover. 

 

Understood. "Period" alignment will avoid notches in the frequency response, but the transient response won't align. According to the paper that I cited (and others), though, as long as they're aligned within a small number of milliseconds (the exact number differs depending upon whom one believes), it won't matter.

[captainbeefheart]

That's how I feel about it, small number of mS is good enough for your average ear.

 

What about crossover frequencies?  Getting that area to sum to flat is higher up on the pecking order for a designer I would assume. Even if you have <1mS time alignment that doesn't incorporate at the crossover frequencies does it? I don't see how it could, if you are in alignment (not at crossover frequencies), at a crossover frequency the difference in phase shift between drivers, not the phase shift at outputs compared to input of network you will always have the driver peak excursion out of alignment to some higher degree. As Edgar called it transient alignment.  Of course this depends on the type of filters you are using but still I would think in order to actually get good time alignment between all the drivers at all frequencies DSP is the only way it's possible.

 

I am not a speaker designer obviously which is why I ask.

[Chief bonehead]

8 hours ago, Edgar said:

 

Understood. "Period" alignment will avoid notches in the frequency response, but the transient response won't align. According to the paper that I cited (and others), though, as long as they're aligned within a small number of milliseconds (the exact number differs depending upon whom one believes), it won't matter.

Hmmm…..that’s not consistent with my tests. Because you have drivers and their accomplices acting together, they will have their own phase response. Time arrivals will let you zero in on when max energy is occurring. But you must have narrow bandwidth around the crossover point to even closer. but even that will not tell you where in period of that specific wavelength you are at because of the particular phase data of that acoustic device. Furthering tuning is needed and that can be define by an etc over a narrowband wide around the crossover point. AND max addition at crossover point. 

[Edgar]

7 hours ago, Chief bonehead said:

Because you have drivers and their accomplices acting together, they will have their own phase response.

 

And that is where things get complicated.

 

It's one thing to implement an electrically perfect crossover network. It's yet another to implement an acoustically perfect balancing network (or whatever one chooses to call it). The magnitude and phase responses of a network consisting of resistors, capacitors, and inductors (or of DSP filters) is easy to predict. But drivers' magnitude and phase responses vary wildly, even within a small frequency band.

 

There are (at least) two ways to deal with this:

1. Incorporate the driver responses into the design. So, for example, if one wants a 2nd-order rolloff, and the driver itself exhibits a 1st-order acoustical rolloff, then all that is needed is a 1st-order electrical rolloff.

2. Equalize the driver responses to flatness, so that the final acoustical response is controlled entirely by the electrical response of the crossover network.

 

Of the two, #1 is easier to implement. #2 is far more difficult, but holds the promise of better ultimate results. DSP is making #2 feasible. The hard part is modeling the driver responses as transfer functions, so that equalization can be better than by trial-and-error.

[Chief bonehead]

15 hours ago, Edgar said:

 

And that is where things get complicated.

 

It's one thing to implement an electrically perfect crossover network. It's yet another to implement an acoustically perfect balancing network (or whatever one chooses to call it). The magnitude and phase responses of a network consisting of resistors, capacitors, and inductors (or of DSP filters) is easy to predict. But drivers' magnitude and phase responses vary wildly, even within a small frequency band.

 

There are (at least) two ways to deal with this:

1. Incorporate the driver responses into the design. So, for example, if one wants a 2nd-order rolloff, and the driver itself exhibits a 1st-order acoustical rolloff, then all that is needed is a 1st-order electrical rolloff.

2. Equalize the driver responses to flatness, so that the final acoustical response is controlled entirely by the electrical response of the crossover network.

 

Of the two, #1 is easier to implement. #2 is far more difficult, but holds the promise of better ultimate results. DSP is making #2 feasible. The hard part is modeling the driver responses as transfer functions, so that equalization can be better than by trial-and-error.

Well….I don’t model. I use actual curves and responses in my programs and designs. I have all the cool toys….

[Edgar]

8 hours ago, Chief bonehead said:

Well….I don’t model. I use actual curves and responses in my programs and designs. I have all the cool toys….

 

I use actual curves to create the models. Nice thing about that is that it allows me to invert transfer functions mathematically, instead of by trial-and-error. (I have some cool toys, too!)

[Tony Whitlow]

Greetings everyone, I have a pair of LA Scala ll & can anyone tell me what would be the measurement from the outside of the cabinet to the rear of the woofer? I'd like to time align my set up.
Thanks in advance! Dan
 

Time alignment is nonsense. They are perfectly fine as they are


Sent from my iPhone using Tapatalk

[VDS]

On 2/28/2022 at 11:28 AM, Chief bonehead said:

My blind tests proved otherwise. 

A 3.5 ms delay between drivers was very clear to me, I especially heard it in the frequency sweep. Listening critically to the sweep is a major tool for me to judge my EQ in Xilica. For right or wrong