Do you mean subwoofer group delay? I haven't paid much attention to it because of other complications.
Assuming the ear+brain need at least 1-2 cycles to detect low freq that's 30ms->60ms @30Hz. If the walls are 5m away the reflection occurs in ~15ms. If multiple subs are used then where is the delay measured from?
Thanks for all the replies! 🙂 I appreciate the references to other sources and works.
Someone mentioned old/dated research. This is one reason why I asked for recent research. I'm familiar with the works of Olson, Knudsen, Harris, Keele, Long, AES and ASA papers, and others, but all of these are from "a while ago".
Not to slight anyone else, but I'd like to thank "kenderes" for the link to the Finland paper. I've accessed it, and it looks very useful, so I'll read it tonight.
Also, "phase_accurate": I thought my inquiry was clear, but I can elaborate. I'm interested in the audible consequences of various crossover designs with regard to phase characteristics. For example, should I care if the high pass and low pass are 90 degrees out of phase? Should I care if the tweeter is 3 inches ahead of the woofer?
Surely, with all the millions of words spoken about these topics, there must be some modern studies and empirical data somewhere.
By the way, phase_accurate, where are you in Switzerland? I'm reasonably familiar with Valais and Vaud, and a little with the area of Visp, Brig, Morel, and Zermatt. I know almost nothing about Basel and Berne. Suisse is my favorite place on Earth.
Someone mentioned old/dated research. This is one reason why I asked for recent research. I'm familiar with the works of Olson, Knudsen, Harris, Keele, Long, AES and ASA papers, and others, but all of these are from "a while ago".
Not to slight anyone else, but I'd like to thank "kenderes" for the link to the Finland paper. I've accessed it, and it looks very useful, so I'll read it tonight.
Also, "phase_accurate": I thought my inquiry was clear, but I can elaborate. I'm interested in the audible consequences of various crossover designs with regard to phase characteristics. For example, should I care if the high pass and low pass are 90 degrees out of phase? Should I care if the tweeter is 3 inches ahead of the woofer?
Surely, with all the millions of words spoken about these topics, there must be some modern studies and empirical data somewhere.
By the way, phase_accurate, where are you in Switzerland? I'm reasonably familiar with Valais and Vaud, and a little with the area of Visp, Brig, Morel, and Zermatt. I know almost nothing about Basel and Berne. Suisse is my favorite place on Earth.
This is a key point. It becomes more noticeable on systems capable of live dynamics. Listen to a well recorded drum kit, slap bass guitar, symphony, acoustic, etc. on a large system outdoors and compare.
I play bass guitar and the difference is very noticeable in the impact coming from a linear phase 3-way PA stack, and the same stack with a typical LR24 implementation. Typical crossovers sound "over processed" and less coherent in comparison.
Hi Dave
I asked the question because it was not clear whether you were talking about the influence of the inter-driver phase relationships on frequency response or temportal response. Well they influence both ! And sometimes even the ones with greater phase-differences can behave better in one discipline than the in-phase cases. A BW 3 crossover for instance has 90 Degrees of phase difference at the crossover frequency compared to LR4 with no phase difference at all. The lobing properties of BW 3 are poor compared to LR4. But it trumps at overall power response and its group delay distortion is also lower.
To go one step futher: The constant voltage crossover that Rod Eliott dislikes is even more out of phase at the crossover frequency but its temporal response is perfect. I once made experiments with a Manger driver and an Audiotechnology Woofer and a modified version of such a crossover. Theresults can be sen here:
Manger
The Tiny step at 4 ms is floor bounce and the "nervousness" right after he initial rise is due to the Manger's ragged frequency response. When this is corrected the step response does look even better. The phase difference at crossover is around 120 Degrees BTW !
Regards
Charles
P.S.: I live in the countryside not far from Berne.
I asked the question because it was not clear whether you were talking about the influence of the inter-driver phase relationships on frequency response or temportal response. Well they influence both ! And sometimes even the ones with greater phase-differences can behave better in one discipline than the in-phase cases. A BW 3 crossover for instance has 90 Degrees of phase difference at the crossover frequency compared to LR4 with no phase difference at all. The lobing properties of BW 3 are poor compared to LR4. But it trumps at overall power response and its group delay distortion is also lower.
To go one step futher: The constant voltage crossover that Rod Eliott dislikes is even more out of phase at the crossover frequency but its temporal response is perfect. I once made experiments with a Manger driver and an Audiotechnology Woofer and a modified version of such a crossover. Theresults can be sen here:
Manger
The Tiny step at 4 ms is floor bounce and the "nervousness" right after he initial rise is due to the Manger's ragged frequency response. When this is corrected the step response does look even better. The phase difference at crossover is around 120 Degrees BTW !
Regards
Charles
P.S.: I live in the countryside not far from Berne.
Last edited:
They are different. They might be better than LR4 in some cases. They are also consistent with frequency.
Not specifically subwoofers. As mentioned in Rods article not much is know about it below 500Hz and it may account for the perception of "slow bass". BTW Earl Geddes assumes about 5 cycles IIRC.
This is exactly where up-to-date research needs to be conducted imo....
With equipment capable of high dynamics, and in an environment that takes the room out of play.
I agree, it's often a very noticeable change in impact (even back indoors).
Heck, sometimes it sounds/feels like certain tunes can drive nails or rip air in two 😀
Yeah, it takes a very large horn system to reproduce a live drum kit and why 'we' only get very compressed recordings: 20,000 Watt Home Hi-Fi System
GM
GM
Good point GM, that high dynamic source material needs to be part to the test spec.
Less compression in recordings definitely helps with my tests comparing audibility of flat phase.
Like ErnieM said, we need live music dynamics....
Less compression in recordings definitely helps with my tests comparing audibility of flat phase.
Like ErnieM said, we need live music dynamics....
The differences I hear between time aligned and phase aligned systems is a "3-D" effect to the playback. Also, the recorded tracks have more separation.
To me, a 3rd order butterworth sounds fine but lacks the effects I mentioned. Anybody else notice this?
To me, a 3rd order butterworth sounds fine but lacks the effects I mentioned. Anybody else notice this?
I'll go with the experts, as quoted above variously, that in normal usage group delay or phase distortion is inaudible if the delays are fixed, relative to other frequencies, not varying with time. This makes sense from a survival standpoint: Animal hearing (including our own) needs to be able to tell the direction of a sound and also if the pitch is varying (comb filtering), estimate probable motions, e.g. of a threat. Evolution is not so concerned that the HF you are listening to right now is advanced a few microseconds compared ot the mids 🙂
Last edited:
An much improved third order crossover,
author Francis Brooke
https://sites.google.com/site/francisaudio69/5-l-enceinte-acoustique/5-3-le-filtrage-des-hp#TOC-5.3.3-LES-FILTRES-QUASI-OPTIMAUX
Formula for 1 kHz (-5.9 dB)
Tweeter :
- inverted
- delay 210 µS (72 mm)
- order 1 at 976 Hz (-3.0 dB)
+ order 2 at 1 kHz (-3.0 dB) , Q=0.707
Woofer :
- order 1, 1.025 kHz (-3.0 dB),
+ order 2, 1kHz (-3.0 dB), Q0.707
Performances :
response variation : 0.82 dB
group delay variation : 177 µS
phase déviation (lo~hi):
490 Hz -54°
500 Hz -51°
1 kHz -13°
2 kHz +48°
2.06 kHz +67°
author Francis Brooke
https://sites.google.com/site/francisaudio69/5-l-enceinte-acoustique/5-3-le-filtrage-des-hp#TOC-5.3.3-LES-FILTRES-QUASI-OPTIMAUX
Formula for 1 kHz (-5.9 dB)
Tweeter :
- inverted
- delay 210 µS (72 mm)
- order 1 at 976 Hz (-3.0 dB)
+ order 2 at 1 kHz (-3.0 dB) , Q=0.707
Woofer :
- order 1, 1.025 kHz (-3.0 dB),
+ order 2, 1kHz (-3.0 dB), Q0.707
Performances :
response variation : 0.82 dB
group delay variation : 177 µS
phase déviation (lo~hi):
490 Hz -54°
500 Hz -51°
1 kHz -13°
2 kHz +48°
2.06 kHz +67°
Last edited:
There are several aspecs to the subject:
A transmission channel with less group-delay distortion has a higher channel capacitiy than a dispersive one.
Regarding the localisation one can reasonably assume that our hearign can more easily establish the interaural delay if all spectral parts of an transient acoustic stimulus arrive at the same time. Temporal dispersion enlarges the size of a perceived phantom source.
Regarding dynamics: I remember that I once read in a book (was it Colloms ?) that the perceived loudness ratios between transient and more "static" signals is affected by group-delay distortion.
But I guess that the transient behaviour doesn't need to be perfect, but how imperfect it may be while not deteriorating sound quality is still not fully determined.
Regards
Charles
A transmission channel with less group-delay distortion has a higher channel capacitiy than a dispersive one.
Regarding the localisation one can reasonably assume that our hearign can more easily establish the interaural delay if all spectral parts of an transient acoustic stimulus arrive at the same time. Temporal dispersion enlarges the size of a perceived phantom source.
Regarding dynamics: I remember that I once read in a book (was it Colloms ?) that the perceived loudness ratios between transient and more "static" signals is affected by group-delay distortion.
But I guess that the transient behaviour doesn't need to be perfect, but how imperfect it may be while not deteriorating sound quality is still not fully determined.
Regards
Charles
What do you mean by that ? That there is no group delay below 500 Hz. It will definitley have group delay below 500 Hz. The group delay is highest at low frequencies and dropping towards higher frequencies.
The crossover that I currently use has more group delay (290 us for 1 kHz crossover frequency) than this one but still less than an LR4 and it is veeeery gently sloping down towards higher frequencies. And both drivers are connected with the same polarity !!!
Regards
Charles
The crossover that I currently use has more group delay (290 us for 1 kHz crossover frequency) than this one but still less than an LR4 and it is veeeery gently sloping down towards higher frequencies. And both drivers are connected with the same polarity !!!
Regards
Charles
With FIR processing being commonly available (any Win10 device will do it, easily), I think we could probably put together some kind of test.
Perhaps even a blind test: here are two IR files. Load them into EQ APO (or whichever software you like), and report back your observations. One will be linear, the other will have some phase shifts.
Chris
Perhaps even a blind test: here are two IR files. Load them into EQ APO (or whichever software you like), and report back your observations. One will be linear, the other will have some phase shifts.
Chris
I meant nothing shown in those results, and there seems to be no audibility study, no doubt it exists.
If there is audible difference with and without phase linearization of any system in any environment shouldn't it be audible if recorded with a mic as well? Could be interesting to hear such recording, if someone had capability and time to make such recording available! 🙂
Last edited:
Yes, information on such things is quite scarce. I remember that there were once some figures that came from Studer professional but I will have to search quite a bit for them.
This is a good idea. But please use group delay patterns which are typical for analog crossovers and not like Blauert and Laws who used used a group delay that was of the peaking narrow band type.
Regards
Charles
Last edited: