• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Dr. Klaus Heinz of HEDD Audio (ex ADAM Audio) - measuring speakers, in particular speaker dynamics

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
Secondly, a closed box can be stuffed more densely, more effectively absorbing the woofer’s back wave and potentially thus improving its transient behaviour by reducing interference on the cone’s motion from internally reflected waves. (Well, you can stuff a ported box very densely if you like, but the port will stop functioning beyond a certain point.) This is primarily a mid-frequency phenomenon.
This would show up in distortion measurements?
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
This would show up in distortion measurements?

Most likely to be visible in the impulse response, and if severe enough as raggedness in the frequency response.

I can’t imagine any obvious mechanism by which it could manifest as THD/IMD.
 

FrantzM

Major Contributor
Forum Donor
Joined
Mar 12, 2016
Messages
4,337
Likes
7,731
Hi

Ignoramus Anonymous:

I am an EE , yet don't understand the notion of filter "ringing", both "pre" and "post" ringings. Can someone explain these to me and why those are issue in filter designs and/or loudspeaker ?
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Hi

Ignoramus Anonymous:

I am an EE , yet don't understand the notion of filter "ringing", both "pre" and "post" ringings. Can someone explain these to me and why those are issue in filter designs and/or loudspeaker ?

Not fully ;) But here's an attempt at a simple explanation...

The function of a crossover is to filter the signal into different frequency bands. Any time a signal is filtered, ringing is produced as a byproduct.

In the case of conventional analogue filters and digital IIR filters (i.e. minimum phase filters), both non-constant group delay (the delaying of some frequencies in relation to others) and ringing result. Lower frequencies arrive later than higher frequencies (non-constant group delay), and ringing occurs after the initial rise of the signal (post-ringing).

The magnitude of both phenomena is dependent upon and proportional to the steepness of the filter slope (steeper filter = more ringing and group delay).

However, non-constant group delay can be avoided by use of a linear phase FIR filter. With such a filter, all frequencies can be made to pass through the filter simultaneously, with no relative delay between different frequencies.

But there is a cost: For the FIR filter to avoid non-constant group delay, it must create not only post-ringing, but also pre-ringing (ringing which occurs prior to the output of the signal).

This graph shows examples of a filter that exhibits only post-ringing (minimum phase, blue) and a filter that exhibits both pre- and post-ringing (linear phase, red):

iu


It's not shown in the above graph, but it can be inferred that the minimum phase filter will also delay low frequencies in relation to high frequencies (non-constant group delay), whereas the linear phase filter will not, i.e. the linear phase filter will allow all frequencies to pass through the filter simultaneously.

It is generally accepted that, above certain thresholds, non-constant group delay, pre-ringing and post-ringing become audible, although the question of where these thresholds lie remains contentious. Various studies have been undertaken, though arguably not enough.

Blauert and Laws investigated the audibility of group delay at various frequencies and concluded that the thresholds were in the range of 1-2ms in the ear's most sensitive region (around 1000Hz IIRC). This corresponds to a phase shift of somewhat more than 360°; referring this to crossover filters, this is equivalent to somewhat more phase shift than that created by a conventional 4th order filter. Subsequent research suggests that these findings are roughly correct, but that it's a bit more complicated, as what is important is not the degree of phase shift/group delay across the whole audio band, but rather within an ERB (which replaces the concept of "critical band"). In other words, the rate of change of group delay is arguably more important than its absolute magnitude; @j_j is the expert on these matters.

Post-ringing is known to be largely masked by the signal as a result of forward masking, and is not of audible concern under normal circumstances (I'm sure exceptions can be concocted, of course).

The question of the audibility of pre-ringing is much more complex. For most people, there is a backward masking effect, such that a louder sound (in this case, the signal) will tend to mask a softer sound occurring up to 20ms prior to it. However, unlike other forms of masking, there is wide subject-to-subject variation in backward masking's duration and degree, with some subjects (particularly with training) exhibiting almost complete immunity to backward masking.

Moreover, IIUC, even in cases where a backward masker renders a sound prior to it in time inaudible, the prior sound may nevertheless affect the perceived loudness of the later/louder sound. In terms of pre-ringing, this implies that it may have a negative effect on the perceived loudness of transients in a signal. This is unfortunately, however, where the extent of my understanding of this topic ends...
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Hi

Ignoramus Anonymous:

I am an EE , yet don't understand the notion of filter "ringing", both "pre" and "post" ringings. Can someone explain these to me and why those are issue in filter designs and/or loudspeaker ?
Actually, there is little to no ringing. :)

The (low-pass) filter is characterized with an impulse. That impulse has infinite bandwidth. When you run it into anything that limits its bandwidth, you get ringing because there is not enough of those individual sine waves to makeup a transients. You can see this in the classic explanation of how to build a square wave out of many sine waves:

Fourier_synthesis_square_wave_animated.gif


Notice the ringing at the transitions and how they get better as we add more high frequency components.

Since music doesn't have infinite bandwidth and doesn't look like an impulse, it is not going to have anywhere close to the ringing we show in classic filter response graphs.
 

Duke

Major Contributor
Audio Company
Forum Donor
Joined
Apr 22, 2016
Messages
1,523
Likes
3,745
Location
Princeton, Texas
I'm kinda late to the party here; was focused on doing an audio show last week.

@Music1969 I tried an experiment comparing a small speaker to a large speaker here: https://audiophilestyle.com/ca/revi...ker-comparison-with-binaural-recordings-r768/...

You can listen to them over headphones to get a good feel for what they sound like.

VERRRY interesting!

I think I hear greater dynamic contrast and greater clarity from the JBLs, as well as better timbre.

@Music1969Another surprise is the audible difference between the two speakers are their directivity indexes. The big speaker, high directivity and the little speakers, wider directivity. So the big speaker has less "room sound" while the smaller, wider directivity speaker had more room sound mixed in with the direct sound.

It is possible that all of the differences I think I hear arise from their respective radiation patterns.

I think the early reflections of the wide-pattern KEFs are degrading the clarity, and I think the higher noise floor generated by the greater reflection density is reducing the dynamic contrast of the KEFs. To my ears the JBLs have more "silence between the notes".
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Not fully ;) But here's an attempt at a simple explanation...

The function of a crossover is to filter the signal into different frequency bands. Any time a signal is filtered, ringing is produced as a byproduct.

In the case of conventional analogue filters and digital IIR filters (i.e. minimum phase filters), both non-constant group delay (the delaying of some frequencies in relation to others) and ringing result. Lower frequencies arrive later than higher frequencies (non-constant group delay), and ringing occurs after the initial rise of the signal (post-ringing).

The magnitude of both phenomena is dependent upon and proportional to the steepness of the filter slope (steeper filter = more ringing and group delay).

However, non-constant group delay can be avoided by use of a linear phase FIR filter. With such a filter, all frequencies can be made to pass through the filter simultaneously, with no relative delay between different frequencies.

But there is a cost: For the FIR filter to avoid non-constant group delay, it must create not only post-ringing, but also pre-ringing (ringing which occurs prior to the output of the signal).

This graph shows examples of a filter that exhibits only post-ringing (minimum phase, blue) and a filter that exhibits both pre- and post-ringing (linear phase, red):

iu


It's not shown in the above graph, but it can be inferred that the minimum phase filter will also delay low frequencies in relation to high frequencies (non-constant group delay), whereas the linear phase filter will not, i.e. the linear phase filter will allow all frequencies to pass through the filter simultaneously.

It is generally accepted that, above certain thresholds, non-constant group delay, pre-ringing and post-ringing become audible, although the question of where these thresholds lie remains contentious. Various studies have been undertaken, though arguably not enough.

Blauert and Laws investigated the audibility of group delay at various frequencies and concluded that the thresholds were in the range of 1-2ms in the ear's most sensitive region (around 1000Hz IIRC). This corresponds to a phase shift of a bit more than 360°, or slightly more phase shift than that created by a conventional 4th order filter. Subsequent research suggests that these findings are roughly correct, but that it's a bit more complicated, as what is important is not the degree of phase shift/group delay across the whole audio band, but rather within an ERB (which replaces the concept of "critical band"). In other words, the rate of change of group delay is arguably more important than its absolute magnitude; @j_j is the expert on these matters.

Post-ringing is known to be largely masked by the signal as a result of forward masking, and is not of concern under normal circumstances (I'm sure exceptions can be concocted, of course).

The question of the audibility of pre-ringing is much more complex. For most people, there is a backward masking effect, such that a louder sound (in this case, the signal) will tend to mask a softer sound occurring up to 20ms prior to it. However, unlike other forms of masking, there is wide subject-to-subject variation in backward masking's duration and degree, with some subjects (particularly with training) exhibiting almost complete immunity to backward masking.

Moreover, IIUC, even in cases where a backward masker renders a sound prior to it in time inaudible, the prior sound may nevertheless affect the perceived loudness of the later/louder sound. In terms of pre-ringing, this implies that it may have a negative effect on the perceived loudness of transients in a signal. This is unfortunately, however, where the extent of my understanding of this topic ends...

Nice info, thank you!

Here is how IR looks with my filters. Pre and post ringing is noticeable in the graph but I can't really say I am able to hear it.

capture.jpg
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Actually, there is little to no ringing. :)

The (low-pass) filter is characterized with an impulse. That impulse has infinite bandwidth. When you run it into anything that limits its bandwidth, you get ringing because there is not enough of those individual sine waves to makeup a transients. You can see this in the classic explanation of how to build a square wave out of many sine waves:

Fourier_synthesis_square_wave_animated.gif


Notice the ringing at the transitions and how they get better as we add more high frequency components.

Since music doesn't have infinite bandwidth and doesn't look like an impulse, it is not going to have anywhere close to the ringing we show in classic filter response graphs.

Question was about pre and post ringing of digital loudspeaker filters, not DAC filters.
 
Last edited:

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Just to clarify, my post earlier in response to @FrantzM was written from the perspective of loudspeaker crossover filters.

I didn’t mean to suggest that there might be audible differences between filters whose ringing and group delay is entirely outside the audio band, as is (hopefully) the case with most DACs.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
Just kidding.. :D Amir said "Since music doesn't have infinite bandwidth ..", but @RayDunzl and @Blumlein 88 demonstrated that sounds produced with 2 spoons and pizza pan with holes practically need infinite bandwidth to reproduce. :D
I must have missed those posts, but it sounds like Amir's music qualification still stands, at least until Ray and Blumlein's 'Things we found in the kitchen' album drops, then all the rules change overnight.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
I must have missed those posts, but it sounds like Amir's music qualification still stands, at least until Ray and Blumlein's 'Things we found in the kitchen' album drops, then all the rules change overnight.

LOOL
Yep, his qualification still stands, but these 2 guys are heavilly trying to shake it! :D
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
the "audible" difference is in the speakers directivity index which determines how much of one's room is mixed in with the direct sound reaching one's ears at the listening position.
More the frequency profile of it. i.e. where the baffle step is. At the top end and the bottom end the size of the baffle isn't going to make much difference. It's in the mid range where it makes a difference - all else being equal.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
I am an EE , yet don't understand the notion of filter "ringing", both "pre" and "post" ringings. Can someone explain these to me and why those are issue in filter designs and/or loudspeaker ?
In a normal linear phase crossover, the ringing of one driver is complementary with the next one, so the ringing cancels out anyway. Of course, the two drivers won't cancel out perfectly at all points in space, and they may be 'voiced' differently, but to a first approximation the ringing is designed to disappear. But people are obsessed with it, anyway, even though they've never heard it...
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
In a normal linear phase crossover, the ringing of one driver is complementary with the next one, so the ringing cancels out anyway. Of course, the two drivers won't cancel out perfectly at all points in space, and they may be 'voiced' differently, but to a first approximation the ringing is designed to disappear.

I should have also mentioned this fact, which I originally learnt from you, thx.

But people are obsessed with it, anyway, even though they've never heard it...

Are you talking about pre-ringing or group delay here? ;)
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
Are you talking about pre-ringing or group delay here?
Pre-ringing. I've never knowingly heard it, although presumably when listening to just, say, the mid range driver in my linear phase speakers I have been subjected to its terrifying consequences. I *have* heard vinyl pre-echo and tape print-through though.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Pre-ringing. I've never knowingly heard it, although presumably when listening to just, say, the mid range driver in my linear phase speakers I have been subjected to its terrifying consequences. I *have* heard vinyl pre-echo and tape print-through though.

And the ringing I posted on that IR? Could that be heard?
 
Top Bottom