Does poor transient response means higher group delay?
Vented enclosure group delay below 100 hz is higher as compared to Sealed enclosure.
1. Does this group delay results in poor transient response? which is why we say Sealed box produces tighter bass whereas Vented box produces muddy bass?
_________________________________________________________
This thread is combined from 5 threads on the same subject.
Vented enclosure group delay below 100 hz is higher as compared to Sealed enclosure.
1. Does this group delay results in poor transient response? which is why we say Sealed box produces tighter bass whereas Vented box produces muddy bass?
_________________________________________________________
This thread is combined from 5 threads on the same subject.
That's been my experience. I'll look for the articles on audibility of group delay--which is what this subject eventually boils down to.
IIRC, there is no real data of audibility of group delay below 500 Hz, although my personal experience is that it is audible, but that might be convolved with other distortion sources. Also IIRC, groups delays less than 3 ms at 500 Hz and below aren't audible (contrary to what Le Cléac'h reported in his horn-compression driver trials--who thought that GD at 0.3 ms was audible: this isn't backed-up by personal experience).
Chris
IIRC, there is no real data of audibility of group delay below 500 Hz, although my personal experience is that it is audible, but that might be convolved with other distortion sources. Also IIRC, groups delays less than 3 ms at 500 Hz and below aren't audible (contrary to what Le Cléac'h reported in his horn-compression driver trials--who thought that GD at 0.3 ms was audible: this isn't backed-up by personal experience).
Chris
Last edited:
See: http://www.diyaudio.com/forums/multi-way/14274-audibility-group-delay.html
and more importantly (I think): Phase, Time and Distortion in Loudspeakers
and more importantly (I think): Phase, Time and Distortion in Loudspeakers
I did a listening trial (not blind) of vented bass boxes (1, 2, and 4 woofers in vented boxes) vs. horn-loaded (horn-loaded driver area equal to one vented woofer area). I can tell you that there is a sharp audibility difference of these two types of bass bin distortions with the horn-loaded by far coming out on top in terms of neutral and convincing sound, but decreasing in audibility as the woofer area grows from 1-->4 woofers. When you look at relative group delay of a big vented box vs. horn loading at the same on-axis SPL, you'll also see a similar difference.
Recall that well-done horn loading of woofers reduces required peak cone motion by about a factor of 5 over the same driver being used in a vented box to produce the same on-axis SPL. Mass effects result in phase lag, and the slope of the phase lag curve vs. frequency is the group delay curve. I've definitely found a correlation between peak motion and audibility of bass distortion, and it seems to follow the group delay curve for vented boxes, but not for sealed direct-radiating boxes.
YMMV.
Chris
Recall that well-done horn loading of woofers reduces required peak cone motion by about a factor of 5 over the same driver being used in a vented box to produce the same on-axis SPL. Mass effects result in phase lag, and the slope of the phase lag curve vs. frequency is the group delay curve. I've definitely found a correlation between peak motion and audibility of bass distortion, and it seems to follow the group delay curve for vented boxes, but not for sealed direct-radiating boxes.
YMMV.
Chris
I did a quick calculation of the Blauert & Laws GD audibility threshold table and found that the wavelength shift for audibility is about 1.6 wavelengths at 500 Hz and 6 wavelengths at 4 kHz. If that trend continues below 500 Hz (human hearing audibility of GD, that is, which is a very tenuous thing to predict), then it would seem that you would need at least a one wavelength shift (lag) at 100 Hz, etc. to be audible. Probably a lot more, however.
This is far more than 1/4 wavelength, which is my point.
This is far more than 1/4 wavelength, which is my point.
Not sure if this is what you mean, but much of what I hear described as "poor transient response," by non-scientists, I would attribute to bad bass/room integration as opposed to any group-delay issues.
Meaning, severe nulls and ringing. The lower speakers attempt to go, the more likely they are to get into problems with this.
In the case of a subwoofer, the integration is made worse by it's decent into the lowest octave as well as having to integrate with a satellite.
This is often why you'll see audiophiles (myself included) use smaller 2-way speakers instead of a 3-way. The claim, that the 2-way sounds more musical, is correct, but not because of a deficiency in the driver, but because they stay away from the monsters of the deep.
Sometimes you'll also see audiophiles / speaker designers claim that an array of smaller speakers is going to be "faster" than one large speaker. Usually bunk. What IS happening is that the array is bass limited.
Other places where you will (IMHO) incorrectly attribute transient response to driver quality is when matching panels to subs. Properly integrated a large diameter sub, ported or sealed, will work marvelously with any panel.
Personally, I would not design a 3-way anymore unless it was for a very well treated room OR I planned on using DSP in the woofer to overcome some of the issues I will invariably face. So I guess this means that when it comes to the bottom octaves I'm with VanDersteen's approach of using powered woofers with EQ, though I'd do it all digitally.
Best,
Erik
Meaning, severe nulls and ringing. The lower speakers attempt to go, the more likely they are to get into problems with this.
In the case of a subwoofer, the integration is made worse by it's decent into the lowest octave as well as having to integrate with a satellite.
This is often why you'll see audiophiles (myself included) use smaller 2-way speakers instead of a 3-way. The claim, that the 2-way sounds more musical, is correct, but not because of a deficiency in the driver, but because they stay away from the monsters of the deep.
Sometimes you'll also see audiophiles / speaker designers claim that an array of smaller speakers is going to be "faster" than one large speaker. Usually bunk. What IS happening is that the array is bass limited. Other places where you will (IMHO) incorrectly attribute transient response to driver quality is when matching panels to subs. Properly integrated a large diameter sub, ported or sealed, will work marvelously with any panel.
Personally, I would not design a 3-way anymore unless it was for a very well treated room OR I planned on using DSP in the woofer to overcome some of the issues I will invariably face. So I guess this means that when it comes to the bottom octaves I'm with VanDersteen's approach of using powered woofers with EQ, though I'd do it all digitally.
Best,
Erik
Last edited:
Thanks!
Was just a thought coming into my head. Didn't calculate anything since at this time of this day I'm far too drunk to calculate anything. ;-)
Merry Christmas and Cheers!
Some prominent figures/designers of audiophile gear quoted arbitrary values of 30ms, 50ms, etc. Others said 1/2 cycle. I've had a hard time believing numbers were derived from any research, because there is no supporting documentation or references with their claims. A single cycle at 100Hz is 10ms duration, and by their methods it was thought that 5ms was acceptable at that point.
It appears that psychoacoustic handling of phase is more precise than that, and it's non-linear. This means it's more sensitive at some frequencies and less so at others. The neurons fire not in time with the frequency, but instead with the phase of the stimulus. Yamaha is one company that has conducted a lot of scientific research into auditory perception and subjective preference. Among their studies was group delay. They determined that a value of about 1.5-2.5ms at 100 Hz was the maximum allowable delay at that frequency.
Phase shift and phase delay are usually one and the same thing, as presented by Bruel & Kjaer in their application note "Loudspeaker Phase Measurements Transient Response and Audible Quality” by Henning Moller, on page 2. As the delay increases, those frequencies are no longer in sync with other frequency components, such as higher harmonics. By definition, this changes the timbre, ie tonal quality of the sounds being reproduced.
There is a post-doctoral research study on group delay below 100 Hz underway right now. It is being conducted by a member of the Atlantic Audio Association and Canadian Acoustical Association. Below, some older papers also hold some information though perhaps outdated:
It appears that psychoacoustic handling of phase is more precise than that, and it's non-linear. This means it's more sensitive at some frequencies and less so at others. The neurons fire not in time with the frequency, but instead with the phase of the stimulus. Yamaha is one company that has conducted a lot of scientific research into auditory perception and subjective preference. Among their studies was group delay. They determined that a value of about 1.5-2.5ms at 100 Hz was the maximum allowable delay at that frequency.
Phase shift and phase delay are usually one and the same thing, as presented by Bruel & Kjaer in their application note "Loudspeaker Phase Measurements Transient Response and Audible Quality” by Henning Moller, on page 2. As the delay increases, those frequencies are no longer in sync with other frequency components, such as higher harmonics. By definition, this changes the timbre, ie tonal quality of the sounds being reproduced.
There is a post-doctoral research study on group delay below 100 Hz underway right now. It is being conducted by a member of the Atlantic Audio Association and Canadian Acoustical Association. Below, some older papers also hold some information though perhaps outdated:
- Determination of loudspeaker signal arrival times: Parts 1,2 and 3. By Richard C. Heyser 1971
- Threshold of Phase Detection by Hearing, Paper of the 44th AES Convention, Rotterdam 1973
Last edited:
Maximum for what...if I might ask? Is there a paper available describing the study?
Chris
Some observations on Group Delay is a good read. Anecdotally, I came to the same conclusion. Seems to be not a lot of research in this area from what I could find.
+1, yes good article.
Often a question of how information is displayed to reveal meaning. The abstract words to this point don't nicely display the sound of group delay.
In the case of the euphemism "group delay", what you really want to see is a 'scope trace of speakers with different group delay characteristcs and esp. if there's somebody who can wipe out the completion by displaying a motional feedback trace. With pulse stimuli or tone bursts, it is easy to decide the kind of cone behaviour you like. Bad group delay is worth seeing on a 'scope.
Bl, Q, or group delay are not parameters you play with in order to have good FR on a sim. They are quality issues you want to control from the start and address the FR by other means.
Ben
Often a question of how information is displayed to reveal meaning. The abstract words to this point don't nicely display the sound of group delay.
In the case of the euphemism "group delay", what you really want to see is a 'scope trace of speakers with different group delay characteristcs and esp. if there's somebody who can wipe out the completion by displaying a motional feedback trace. With pulse stimuli or tone bursts, it is easy to decide the kind of cone behaviour you like. Bad group delay is worth seeing on a 'scope.
Bl, Q, or group delay are not parameters you play with in order to have good FR on a sim. They are quality issues you want to control from the start and address the FR by other means.
Ben
Last edited:
It is important to note that group-delay as such is not a problem. It is the group-delay distortion that is.
I guess our susceptibility to group-delay distortion has been determined by evolution and it is therefore highest where it makes most sense.
If a transient signal arrives at our ears the difference in arrival time between our two ears is used for lateralisation (I.e. the determination of the direction where it is coming from). Transient signals are the ones that can be localised best while a constant sinusoidal is the hardest to localise. Because it is physically and mathematically difficult to accurately determine a delay that is only a small portion of a signal period, our perception of direction is less acute at low frequencies and therefore the susceptibility to group delay distortion is lower down there.
At higher frequencies the HRTF do reduce the level of the signal arriving at the opposite ear. Therefore the lateralisation at higher frequencies is done via intensity differences rather than arrival time. And I am convinced that is the reason why we are less susceptible to group-delay distortion higher up than we are the midrange frequencies like shown by Bluert and Laws. Even though I suspect that their result might lack some accuracy.
Group-delay distortion does temporally smear transient signals depending on frequency. This has some effect on lateralisation. It does not change the actual direction where it is perceived as coming from (given that the group delay distortion is the same for both channnels). But it results in lateral smearing of the perception of direction. I.e. a lateral widening of the perceived source is taking place.
The other effect of group delay distortion is simply taking the temporal precision from transient signals like percussion instruments.
In normal living rooms the transient response of a loudspeaker - especially at low frequencies - is often swamped by the transient response of the room. But there is a tendency to more accurate transient behaviour of closed boxes that don't expand too low. If the room permits, a bass-reflex speaker that goes significantly (!) lower can have an equally nice transient response at those frequencies where it actually matters. Keep in mind that musical instruments that can really go low, like organ for instance, don't sound very "impulsive" down there while instruments like drums are usually located a little higher in spectrum.
Regards
Charles
I guess our susceptibility to group-delay distortion has been determined by evolution and it is therefore highest where it makes most sense.
If a transient signal arrives at our ears the difference in arrival time between our two ears is used for lateralisation (I.e. the determination of the direction where it is coming from). Transient signals are the ones that can be localised best while a constant sinusoidal is the hardest to localise. Because it is physically and mathematically difficult to accurately determine a delay that is only a small portion of a signal period, our perception of direction is less acute at low frequencies and therefore the susceptibility to group delay distortion is lower down there.
At higher frequencies the HRTF do reduce the level of the signal arriving at the opposite ear. Therefore the lateralisation at higher frequencies is done via intensity differences rather than arrival time. And I am convinced that is the reason why we are less susceptible to group-delay distortion higher up than we are the midrange frequencies like shown by Bluert and Laws. Even though I suspect that their result might lack some accuracy.
Group-delay distortion does temporally smear transient signals depending on frequency. This has some effect on lateralisation. It does not change the actual direction where it is perceived as coming from (given that the group delay distortion is the same for both channnels). But it results in lateral smearing of the perception of direction. I.e. a lateral widening of the perceived source is taking place.
The other effect of group delay distortion is simply taking the temporal precision from transient signals like percussion instruments.
In normal living rooms the transient response of a loudspeaker - especially at low frequencies - is often swamped by the transient response of the room. But there is a tendency to more accurate transient behaviour of closed boxes that don't expand too low. If the room permits, a bass-reflex speaker that goes significantly (!) lower can have an equally nice transient response at those frequencies where it actually matters. Keep in mind that musical instruments that can really go low, like organ for instance, don't sound very "impulsive" down there while instruments like drums are usually located a little higher in spectrum.
Regards
Charles
Last edited:
Hi Charles!
Excellent and interesting post yet again the only thing I would add is that when it comes to drums (or bass guitar or upright) the reason they sound more impulsive than an organ pipe is IME high frequency content in the kHz region. If you were to eq that away and leave the drums fundamental the result sounds very much like an organ pipe and not at all 'impulsive'.
As far as I can tell transients are something that happens in the high-mid and treble region while speaker-related group delay is a LF phenomenon. I may have a different/wrong understanding of transient though, I take it to be the same as 'attack'.
Excellent and interesting post yet again the only thing I would add is that when it comes to drums (or bass guitar or upright) the reason they sound more impulsive than an organ pipe is IME high frequency content in the kHz region. If you were to eq that away and leave the drums fundamental the result sounds very much like an organ pipe and not at all 'impulsive'.
As far as I can tell transients are something that happens in the high-mid and treble region while speaker-related group delay is a LF phenomenon. I may have a different/wrong understanding of transient though, I take it to be the same as 'attack'.
Hi Charles
The statement about the higher frequency content is of course true ! There is one famous Sheffield Lab recording with an equally famous Japanese drum ensemble (Kodo). It has one track where a drum of about 160 cm diameter is playing the main role. I do sometimes use this one for demoeing how much of that giant drum's attack is lost when the tweeter is covered up.
But I would not dare to name an exact figure percentage how much of the transient accuracy/perception is delivered by the amplitude response alone. But I would say it is the bigger part. But time-smear due to group delay distortion does also play a role when it comes to the reduction of transient accuracy. High-Q resonances do also play a role in this respect (and do also increase group-delay distortion).
It is actually true that the harmonics will start first on a bass string for instance and the fundamental is the slowest one in terms of "buildup" (don't know a better term). The lower the string in frequency the more it is like that. That's the reason why some bass players use a baritone tuning (i.E. ADGC instead of EADG) for fast and/or percussive (slapping & popping etc) solo works - the higher tuned strings simply respond faster.
And yes: I always smile when I hear the term "fast bass".
Regards
Charles
The statement about the higher frequency content is of course true ! There is one famous Sheffield Lab recording with an equally famous Japanese drum ensemble (Kodo). It has one track where a drum of about 160 cm diameter is playing the main role. I do sometimes use this one for demoeing how much of that giant drum's attack is lost when the tweeter is covered up.
But I would not dare to name an exact figure percentage how much of the transient accuracy/perception is delivered by the amplitude response alone. But I would say it is the bigger part. But time-smear due to group delay distortion does also play a role when it comes to the reduction of transient accuracy. High-Q resonances do also play a role in this respect (and do also increase group-delay distortion).
It is actually true that the harmonics will start first on a bass string for instance and the fundamental is the slowest one in terms of "buildup" (don't know a better term). The lower the string in frequency the more it is like that. That's the reason why some bass players use a baritone tuning (i.E. ADGC instead of EADG) for fast and/or percussive (slapping & popping etc) solo works - the higher tuned strings simply respond faster.
And yes: I always smile when I hear the term "fast bass".
Regards
Charles
That's right. The 'felt' or audible impulsive character is mainly created by the impulse response, that's why even at a bassdrum or a snaredrum the tweeter plays a big part of the sound impression since it forms the initial response or, in other words, the attack of the signal since the LF transducers are much slower in the transient response.
That's not the same. Like explained above, the woofer or subwoofer does not form the initial, first audible part of the impulse, which is the most important thing about how the impulse response is perceived. But that does not only happen in the LF, you will also experience the same problem with a (long) midrange horn if it's not geometrically corrected mounted distance of the source of the sound, compared to the tweeter. It's just not that common or extreme with the usual direct radiating mid- and highrange-trancducers.
With a DSP you can eliminate most of the group delay via delay and/or phase correction but you'll notice in the bass it can't do a complete and perfect compensation. For the audible impression there are three things which can't be corrected that way.
- The more extended low frequency the speaker can reproduce the slower it appears to be regarding impulse response. That's true and a physical fact because the signal takes much longer to go through the system because of the wavelength. You can improve it by cutting off lower frequencies (low cut/subsonic) but that ofcourse means the speaker can't reproduce that deep frequencies anymore. There's no solution for that without cutting off some part of the spectrum.
- The DSP (or an equalizer) can't correct the response of the room at every and any location in the room. That means you will have dips and (worse for the perception of the precision) peaks at the resonance frequencies of the room. That can be improved a lot by using a SBA or DBA (single/double bass array).
- The DSP can't eliminate the decay time of the resonances in the room, of the speaker and the port/line/horn/whatever. The decay time of the loudspeaker is massively increased at the fs and the port frequency, you can't reduce that. This plays a huge part in why sealed enclosures are precieved as 'cleaner' and 'more precise' than vented loudspeakers, they simply don't have the huge decay time of the 2nd resonance of the port - which often in turn then initiate the room resonances too.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.