What do group delay/step response reflect?

[DNCAgain]

I hate that I keep asking this, but I'm still unclear: what exactly do phase distortion, group delay, and step response reflect? If their impact is minor or they typically carry little information, why does Amir insist on testing them?

 

[amirm]

I do them because I got tired of explaining why they are not necessary. No other reason.

 

[DNCAgain]

Also, I've encountered this view for too long: multi-driver headphones and speakers are not minimum-phase systems, so frequency response cannot be fully equated with time domain.

I deeply feel that I lack knowledge. Can somebody recommend a relevant electroacoustics book to me?

 

[DNCAgain]

I do them because I got tired of explaining why they are not necessary. No other reason.

Actually, I think you could compile some of the answers into a read first section, explaining what metrics to test, what not to test, and why.

 

[MaxwellsEq]

I deeply feel that I lack knowledge

Yes, based on how many questions you ask, you appear to be starting from a low base of knowledge polluted by nonsense in magazines and forums.

I can't stress enough how much benefit you would gain by reading something like this:
The Art of Electronics

 

[DNCAgain]

Yes, based on how many questions you ask, you appear to be starting from a low base of knowledge polluted by nonsense in magazines and forums.

I can't stress enough how much benefit you would gain by reading something like this:
The Art of Electronics

Thanks for sharing!

 

[amirm]

Also, I've encountered this view for too long: multi-driver headphones and speakers are not minimum-phase systems, so frequency response cannot be fully equated with time domain.

So?

 

[Keith_W]

So?

You misunderstood him. He isn't arguing for the validity of that viewpoint, he is asking what it means. See his second sentence, "I deeply feel I lack knowledge".

@DNCAgain the first thing you need to do is understand what "minimum phase" means. John Mulcahy (author of REW) has a very nice article about it here. That explanation is a bit technical, and it requires you to concentrate and study. I attempted a more simplified explanation in my REW eBook (see signature for link). Once you understand that, the next thing to understand is that ALL drivers, crossover components, and electronic components are minimum-phase. A single room reflection is also minimum-phase. However, multiple minimum phase systems cascaded on top of each other may or may not result in a minimum-phase system.

Take a room reflection. ONE reflection is minimum-phase. However, that reflection interacts with other room reflections. If the wavelength is very long with respect to room dimensions, it will form predictable patterns in a room. These are minimum-phase regions in the room response. As wavelengths get shorter, at some point the patterns are chaotic and no longer predictable - so it is no longer minimum-phase.

The statement "multi-driver headphones and speakers are not minimum-phase" is mostly untrue. Remember that minimum-phase means that the response and its inversion are causal and stable. Ignore the term "causal" for now, and just focus on "stable". "Stable" means "not trending towards infinity". For example, a null caused by two signals which are equal but 180deg out-of-phase to each other. If you attempt to correct it by inversion, it will require infinite energy to fill. If it's not perfectly 180deg (say 175deg) it will not require infinite energy, merely a lot of energy. But you might still burn up amplifiers and melt voice coils trying to fill that null.

Now, let's talk speakers. A single driver is minimum-phase, a crossover network is minimum-phase. Put the two of them in series, and the result is a convolution (multiplication) of the characteristics of both. A zero multiplied by a nonzero value is zero, so the result is always minimum-phase. But if you have two drivers, or two reflections, or a woofer and a port, the result is a summation of both responses. A zero added to a nonzero value results in a nonzero value, so the result may or may not be minimum-phase.

If you have an incompetently designed loudspeaker where the crossover point between (say) mid and tweeter is exactly 180deg out of phase, this will result in a speaker which is non minimum-phase. Fortunately, most speakers are not like that, save for the most idiotic designs or some DIY hackjobs. The vast majority of speakers ARE minimum-phase.

Sadly I can not think of a good electroacoustics book. The closest I can think of is Vance Dickason's Loudspeaker Design Cookbook which is filled with valuable information about loudspeakers. Some people will recommend Toole but IMO its discussion on the specifics of loudspeaker engineering is a bit lacking, its focus is on other things. But you still need that book though.

 

[DNCAgain]

You misunderstood him. He isn't arguing for the validity of that viewpoint, he is asking what it means. See his second sentence, "I deeply feel I lack knowledge".

@DNCAgain the first thing you need to do is understand what "minimum phase" means. John Mulcahy (author of REW) has a very nice article about it here. That explanation is a bit technical, and it requires you to concentrate and study. I attempted a more simplified explanation in my REW eBook (see signature for link). Once you understand that, the next thing to understand is that ALL drivers, crossover components, and electronic components are minimum-phase. A single room reflection is also minimum-phase. However, multiple minimum phase systems cascaded on top of each other may or may not result in a minimum-phase system.

Take a room reflection. ONE reflection is minimum-phase. However, that reflection interacts with other room reflections. If the wavelength is very long with respect to room dimensions, it will form predictable patterns in a room. These are minimum-phase regions in the room response. As wavelengths get shorter, at some point the patterns are chaotic and no longer predictable - so it is no longer minimum-phase.

The statement "multi-driver headphones and speakers are not minimum-phase" is mostly untrue. Remember that minimum-phase means that the response and its inversion are causal and stable. Ignore the term "causal" for now, and just focus on "stable". "Stable" means "not trending towards infinity". For example, a null caused by two signals which are equal but 180deg out-of-phase to each other will require infinite energy to fill. If it's not perfectly 180deg (say 175deg) it does not require infinite energy, merely a lot of energy. But you will burn up amplifiers and melt voice coils trying to fill that null.

Now, let's talk speakers. A single driver is minimum-phase, a crossover network is minimum-phase. Put the two of them in series, and the result is a convolution (multiplication) of the characteristics of both. A zero multiplied by a nonzero value is zero, so the result is always minimum-phase. But if you have two drivers, or two reflections, or a woofer and a port, the result is a summation of both responses. A zero added to a nonzero value results in a nonzero value, so the result may or may not be minimum-phase.

If you have an incompetently designed loudspeaker where the crossover point between (say) mid and tweeter is exactly 180deg out of phase, this will result in a speaker which is non minimum-phase. Fortunately, most speakers are not like that, save for the most idiotic designs or some DIY hackjobs. The vast majority of speakers ARE minimum-phase.

Sadly I can not think of a good electroacoustics book. The closest I can think of is Vance Dickason's Loudspeaker Design Cookbook which is filled with valuable information about loudspeakers. Some people will recommend Toole but IMO its discussion on the specifics of loudspeaker engineering is a bit lacking, its focus is on other things. But you still need that book though.

Thank you for explaining that for me, and thank you for sharing!
I think it's time for me to settle down and start studying seriously!：）

 

[Jiraya369]

Also, I've encountered this view for too long: multi-driver headphones and speakers are not minimum-phase systems, so frequency response cannot be fully equated with time domain.

I deeply feel that I lack knowledge. Can somebody recommend a relevant electroacoustics book to me?

it's just bullshit made up by people who don't know, just like you, and refuse to believe the simple fact that FR is all that's needed to evaluate headphones/iems, granted the distortion isn't through the roof (highly unlikely with iems) and comfort is there.

For speakers, since sound bounces around, you need to look at on AND off axis data, not to mention the fact that the room has an effect on the sound below the schroeder frequency so the bass freqs and stuff will have peaks and dips NOT produced by speakers, thus anechoic data is required.

That's pretty much it.

 

[DNCAgain]

it's just bullshit made up by people who don't know, just like you, and refuse to believe the simple fact that FR is all that's needed to evaluate headphones/iems, granted the distortion isn't through the roof (highly unlikely with iems) and comfort is there.

For speakers, since sound bounces around, you need to look at on AND off axis data, not to mention the fact that the room has an effect on the sound below the schroeder frequency so the bass freqs and stuff will have peaks and dips NOT produced by speakers, thus anechoic data is required.

That's pretty much it.

Thanks!