Can someone who is knowledgeable provide me with answers (and importantly evidence) to the significance of phase in multiway speakers?
For example in a 12db/octave crossover, it is common to reverse the phase of the tweeter. What is the audible significance of doing that?
What is the advantage of having a very small acoustic phase deviation between the two drivers?
For example in a 12db/octave crossover, it is common to reverse the phase of the tweeter. What is the audible significance of doing that?
What is the advantage of having a very small acoustic phase deviation between the two drivers?
The original source material is recorded at a single point in space (dual points in the case of stereo). Of course this is an idealised description, as in many recordings multiple microphones are used, and the sound stage may be a complete fictional synthesis.
The object of most reproducer systems is to recreate the original sound stage in pursuit of a realistic effect.
This is comparatively simple (!) to achieve, at least in an idealised system, where there is a single (or dual in the case of stereo) driver.
Where multiple drivers are employed it is not so straightforward to achieve an effect identical to that of the original source, for one reason or another, principally the different distances between the drivers and the listener's ear. For this reason various stratagems are employed in an effort to nullify these effects. The exact nature of these depend on the details of the system in question.
w
Rod Elliott has a couple of pages on this subject:- http://sound.westhost.com/pcmm.htm
The object of most reproducer systems is to recreate the original sound stage in pursuit of a realistic effect.
This is comparatively simple (!) to achieve, at least in an idealised system, where there is a single (or dual in the case of stereo) driver.
Where multiple drivers are employed it is not so straightforward to achieve an effect identical to that of the original source, for one reason or another, principally the different distances between the drivers and the listener's ear. For this reason various stratagems are employed in an effort to nullify these effects. The exact nature of these depend on the details of the system in question.
w
Rod Elliott has a couple of pages on this subject:- http://sound.westhost.com/pcmm.htm
Last edited:
I'm not very knowledgeable about phase, but the example of the 12db crossover is pretty simple.
The filters for each of the low and high pass sections result in a total of 180 degrees phase difference between the woofer and the tweeter at the crossover frequency. Think of this from the perspective of two sine waves arriving at the same point at the same time. Say one of the sinewaves is at its peak at that point in space at a particular time, then the other sinewave will be at its trough at the same point in space at the same time.
Because the waves are (theoretically) exactly the opposite they cancel each other out resulting in a null in the freq response at that particular freq.
By inverting one of the drivers (usually the tweeter) the waves arrive IN phase at the point in space and time and no cancelation happens, instead the waves sum together to provide (depending on the type of filter) something varying from a flat frequency response to a bit of a hump.
Below is a simulation (based on real measurements) of what happens when the tweeter is inverted.
The black trace is the spl trace when the tweeter is inverted, the blue trace is the spl trace that results if the tweeter is NOT inverted.
The speakers in the below crossover sim are Time aligned. It is an accoustic 2nd order bessel crossover. Electrically it is a bit hard to say
a notch filter on the woofer and single cap on the tweeter.
Tony.
Note that the results for this crossover when doing actual measurements of the implemented crossover (with correct phase) were VERY close to the simulation. I did do measurements with the tweeter inverted as well (but did not keep them) and from memory they looked just like the sim shows.
To show that the simulation is accurate, The second graph shows the actual measured response of the speaker with the built crossover (black trace) compared to the value calculated by the simulator for the crossover from the individual driver accoustic and impedance measurements (red trace).
The filters for each of the low and high pass sections result in a total of 180 degrees phase difference between the woofer and the tweeter at the crossover frequency. Think of this from the perspective of two sine waves arriving at the same point at the same time. Say one of the sinewaves is at its peak at that point in space at a particular time, then the other sinewave will be at its trough at the same point in space at the same time.
Because the waves are (theoretically) exactly the opposite they cancel each other out resulting in a null in the freq response at that particular freq.
By inverting one of the drivers (usually the tweeter) the waves arrive IN phase at the point in space and time and no cancelation happens, instead the waves sum together to provide (depending on the type of filter) something varying from a flat frequency response to a bit of a hump.
Below is a simulation (based on real measurements) of what happens when the tweeter is inverted.
The black trace is the spl trace when the tweeter is inverted, the blue trace is the spl trace that results if the tweeter is NOT inverted.
The speakers in the below crossover sim are Time aligned. It is an accoustic 2nd order bessel crossover. Electrically it is a bit hard to say
a notch filter on the woofer and single cap on the tweeter. Tony.
Note that the results for this crossover when doing actual measurements of the implemented crossover (with correct phase) were VERY close to the simulation. I did do measurements with the tweeter inverted as well (but did not keep them) and from memory they looked just like the sim shows.
To show that the simulation is accurate, The second graph shows the actual measured response of the speaker with the built crossover (black trace) compared to the value calculated by the simulator for the crossover from the individual driver accoustic and impedance measurements (red trace).
Attachments
Last edited:
Here is an impeadence and phase plot for a 3 way system with 24db acoustic slopes. As you can see asside from the resonance and the box tuning the actual crossover slopes don't show up in the electrical plot.
As long as you have correct suming through the crossover points I wouldn't loose sleep over it. I know others put quite a bit of emphasis on it as opionins certainly do vary on this topic.
Rob
As long as you have correct suming through the crossover points I wouldn't loose sleep over it. I know others put quite a bit of emphasis on it as opionins certainly do vary on this topic.
Rob
Attachments
Last edited:
try it and listen
Professor smith,
my experience over many years, with many listeners, is that some hear an effect when one driver is connected in reverse Polarity to the other driver, and that other listeners SEEM to not hear any difference regardless of relative Polarities
-{that is when the crossover type chosen does not cause a notch in the frequency response for the polarity used}.
I state SEEM, because for some listeners the issue of relative polarity seems to be a low priority in the mechanism of their particular brains' processing of audio stimulae.
For other listeners, relative polarity is clearly audible, and most can state a preference for one option over the other,
though with recorded performances of human voices and/or musical instruments the preference is sometimes dependant of the audio characteristics of the recording ... that is, it is a different preference for some recordings than for others.
I have found this change of preference occurs when the program sources are popular types of music that have been electronically processed
-{"mixed" as recording engineers call it}- to sound impressive at initial hearing via small radios and low to medium quality home audio systems.
For Classical Music, and other types which are recorded with minimal altering of the waveforms of the signal, the listener preference polarity usually does not change.
Myself, I have always heard an effect which I later discovered was caused by the relative polarities of the 2, or more, drivers in the loudspeaker system.
I could attempt to describe it, but better is you try an experiment and listen to determine whether your brain function has any degree of priority for this.
Simplest is when one has a loudspeaker with either a 1st order or a 3rd order slopes' crossover -{Total of slopes of the drivers' roll-offs plus those of the electrical filters}- because both can be used in either polarity of one driver with respect to the other,
BUT, first find the optimum listening height/vertical listening axis, because the vertical off-axis + and - lobes reverse in direction when the polarity of one driver is changed ... that is very audible if one happens to have been listening on or near to axis of one of the lobes !
Also, reflections from the floor can cause a change in sound when the vertical lobes are reversed.
The sound changes caused by lobes are different in character to the audio characters of different relative polarities, to my hearing at least !
Professor smith,
my experience over many years, with many listeners, is that some hear an effect when one driver is connected in reverse Polarity to the other driver, and that other listeners SEEM to not hear any difference regardless of relative Polarities
-{that is when the crossover type chosen does not cause a notch in the frequency response for the polarity used}.
I state SEEM, because for some listeners the issue of relative polarity seems to be a low priority in the mechanism of their particular brains' processing of audio stimulae.
For other listeners, relative polarity is clearly audible, and most can state a preference for one option over the other,
though with recorded performances of human voices and/or musical instruments the preference is sometimes dependant of the audio characteristics of the recording ... that is, it is a different preference for some recordings than for others.
I have found this change of preference occurs when the program sources are popular types of music that have been electronically processed
-{"mixed" as recording engineers call it}- to sound impressive at initial hearing via small radios and low to medium quality home audio systems.
For Classical Music, and other types which are recorded with minimal altering of the waveforms of the signal, the listener preference polarity usually does not change.
Myself, I have always heard an effect which I later discovered was caused by the relative polarities of the 2, or more, drivers in the loudspeaker system.
I could attempt to describe it, but better is you try an experiment and listen to determine whether your brain function has any degree of priority for this.
Simplest is when one has a loudspeaker with either a 1st order or a 3rd order slopes' crossover -{Total of slopes of the drivers' roll-offs plus those of the electrical filters}- because both can be used in either polarity of one driver with respect to the other,
BUT, first find the optimum listening height/vertical listening axis, because the vertical off-axis + and - lobes reverse in direction when the polarity of one driver is changed ... that is very audible if one happens to have been listening on or near to axis of one of the lobes !
Also, reflections from the floor can cause a change in sound when the vertical lobes are reversed.
The sound changes caused by lobes are different in character to the audio characters of different relative polarities, to my hearing at least !
Last edited:
Once they heard the effect of what truly phase-linearized speakers can do with good recordings, only few people will ever go back to "time-smearing" speakers, I think.
Personally I experience an increase in convincing spaciousness but the most prominent effect for me is the tightness, perceived "correctness", of percussive sounds, especially those with low frequency content. This seems to call for a phase linearization beyond the lower cutoff frequency of the speaker, though (not all people agree that this is required or favorable). Then also absolute polarity can be easily detected, much easier than with a non-corrected or only partly corrected speaker.
When the phase linearization (but no other change) is done with an FIR-filter in the digital domain, applied to a conventional speaker, results can easily be compared (even with true blind-tests). There are changes also to the timbre of sounds, it gets a little darker and extending lower, but also clearer and more distinguishable. IHMO, all that.
Personally I experience an increase in convincing spaciousness but the most prominent effect for me is the tightness, perceived "correctness", of percussive sounds, especially those with low frequency content. This seems to call for a phase linearization beyond the lower cutoff frequency of the speaker, though (not all people agree that this is required or favorable). Then also absolute polarity can be easily detected, much easier than with a non-corrected or only partly corrected speaker.
When the phase linearization (but no other change) is done with an FIR-filter in the digital domain, applied to a conventional speaker, results can easily be compared (even with true blind-tests). There are changes also to the timbre of sounds, it gets a little darker and extending lower, but also clearer and more distinguishable. IHMO, all that.
Last edited:
Just read this or the Master Handbook of Acoustics' first and second chapter. Same result either way.
I wouldn't get caught up in the he said she said end of the question. Until the naysayers (or perhaps the yeasayers in this case) have any evidence, ignore them. The experiment says phase is inaudible with a music signal, but can be audible with test tones. I've asked for evidence wrt phase audibility a few times and no one has produced any.
Until then, I'm staying put.
Dan
I perfectly can accept this result but it's based on an IMHO invalid prerequisite: The music signal shall be constrained to "real" music sources only, established instruments and voices and such.
But in a wider and more approbriate context for today's different music genres, any signal that a good loudspeaker is capable to reproduce without major problems must be taken into account when judging the audibilty of phenomena like absolute polarity, linear phase etc. With artificial sounds one can do only preference/discrimination tests, not asking things like "which one sounds closer to reality?"
I can provide a test signal that in itself is immune against any phase shift of the loudspeaker but clearly illustrated the audible effects of shifting harmonics around in phase. It is not clear though that the perceived shift in timbre is mostly because of nonlinearites of the speaker (to identify this was the design goal at first) or mostly from nonlinearities of the ear itself. My experimentation tends to favor the second as the effect was there even at very low volumes and close distances where the speakers do not contribute significant amounts of distortion. Of course the hard thing is to prove that scientifcally...
Last edited:
It makes a difference. To some people it doesn't matter, other things are more important.
It is all a matter of what compromises make the most sense for YOU. And only you are the final arbiter there.
Those that are proponents find value in phase coherence, those that don't, don't (or have never been properly exposed -- it isn't just the speakers that play a role)
Personally you won't find me giving up my small (treated) FRs (at least for covering the critical range) because i really value what good phase coherence brings me. In conjunction with good downward dynamic range (and a good recording)...
dave
It is all a matter of what compromises make the most sense for YOU. And only you are the final arbiter there.
Those that are proponents find value in phase coherence, those that don't, don't (or have never been properly exposed -- it isn't just the speakers that play a role)
Personally you won't find me giving up my small (treated) FRs (at least for covering the critical range) because i really value what good phase coherence brings me. In conjunction with good downward dynamic range (and a good recording)...
dave
Same for me, basically (I often use phase-corrected small coaxial monitors).
And I doubt that anyone would not be deeply affected by the sonic impact of an AER MD3 in an Oris150 + Bass. To me it was the key incident to be careful about what "a silly whizzer-cone fullranger" can or cannot do, but that's a different topic...
To some people indeed other things are more important and it's all a matter of preferences.
Klaus
Because music don't stop at 50hz and 10khz
That some are happy to live with those restrictions to keep their phase says a lot right their... but...
The new gen2 Mark Audios are perfectly capable of 40-20k+ in the right box and they are just an example of some of the more capable of today's FR (the sadly discontinued Fostex FF85K being a personal favorite -- 150-30k). An XO <400k can be quite innocuous (ie less evil) and gives you lots of latitude.
and similarily a supertweeter up that high is also usually fairly innocuous.
dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.