"...the significance of phase in multiway speakers" is too broad a question. You see the result.
If the intention is to ask what is the audible significance of reversing tweeter polarity in a 12dB/octave crossover - the answer is that it depends on what the individual driver slopes are and the relative positions of the acoustic centers of the drivers. The crossover is the sum of the driver plus crossover slopes (and phase). If the woofer and tweeter have 2nd order high and lowpass and highpass slopes, respectively, the response of the 12dB (electrical) crossover may be (typically) 2nd, 3rd, or 4th order depending on the corner frequency of those slopes and how the crossover is designed.
Lets assume we have a coaxial woofer and tweeter with aligned acoustic centers. Lets further assume they act like pure resistors and there is 2+ octaves of overlap in their response and we cross with a well designed 2nd order crossover in the center of the overlap. We listen with normal polarity and experience a suckout at crossover, we then flip the tweeter polarity and get a flat response. That the tweeter and woofer polarities do not match doesn't matter. A lot of people will try to sell you on preserving waveshapes using words like transient perfect, etc... That the initial tweeter pulse is opposite that of the woofer and the impulse or step response looks odd does not matter. There is no evidence that we can hear waveshapes. The ear is made up of a combined set of bandpass detectors (cilia) that are laid out on a spiral membrane in your cochlea. The ear is sensitive to spectrum (frequency) response, not waveshape.
If the driver slopes are such that a 3rd or 4th order slope is attained with the 2nd order network, inverting polarity may actually cause no change, or a dip, respectively and thus the "correct polarity" depends on crossover and driver details and the "correct" polarity is the one that gives the flattest response. If the tweeter leads the woofer, as in a typical speaker, the correct polarity can flip depending on where the crossover frequency is.
The audibility of absolute phase (speaker system polarity) is a separate issue, and in my opinion has no meaning with test signals or music because you have no control over the recording process and don't know if the phase of the end result of mixing and recording and playback has the correct phase. Furthermore speaker distortion clouds the issue.
Design your system for flat response, and reasonably flat and smooth off axis response. If you wish to design your speaker for preserving waveshapes, that is an engineering (rather than sonic) goal. It is one that is a challenge, but the audible results are more a case of "I did that" than anything else.
Last edited:
I really dont know what it is you guys want
what you do right now is just negative counter posting, turning original posts around, malishing others posts
that way you achieve nothing other than a cramped thread
take care, cheers
see my Post, #8 on Page 1
Sorry Ron E,
some people can hear it.
I explained this case in my earlier post -{Page 1, #8}.
I can remember hearing the differences before I knew what it was,
and before I knew anything about crossover networks or that anything was inside a loudspeaker enclosure other than the backs of the drivers I could see the fronts of.
I hear it at every demonstration of previously unknown to me models of loudspeakers.
To date I have never been mistaken.
Regardless of whether 1st Order in like or unlike Polarity,
or 2nd Order in unlike Polarity,
or 3rd Order in like or unlike Polarity,
or 4th Order in like Polarity,
I can always identify the relative Polarity of one driver versus the other simply by listening to the loudspeaker playing music.
{I cannot always identify the crossover slopes' Order by listening.}
I have never tried listening to Sine Waves, or any other form of Test Signals,
when auditioning loudspeakers.
I think one may need to hear Transient dense program to hear this effect.
I hear it very distinctly with any percussion instrument,
and also with Piano music and any Orchestral that includes Pizzicato playing.
I have met people who can hear it.
Some are recording engineers.
If you don't hear it, consider yourself fortunate, as choosing or designing loudspeakers may be simpler for you than for those of us who do hear it.
*********************************************************
Hello tinitus,
good to read you here, and that you are continuing to trust your hearing !
Problem with these arguments is that it seems some people will not believe anything that they cannot experience themselves.
Human Nature has been this way since the earliest recorded evidence at least, and likely it will remain this way.
Myself, I do not intend to be a fool ... things do happen beyond my perception abilities.
I am posting here simply so that inquisitive people will not give up investigating the types of phenomena which may occur.
Sorry Ron E,
some people can hear it.
I explained this case in my earlier post -{Page 1, #8}.
I can remember hearing the differences before I knew what it was,
and before I knew anything about crossover networks or that anything was inside a loudspeaker enclosure other than the backs of the drivers I could see the fronts of.
I hear it at every demonstration of previously unknown to me models of loudspeakers.
To date I have never been mistaken.
Regardless of whether 1st Order in like or unlike Polarity,
or 2nd Order in unlike Polarity,
or 3rd Order in like or unlike Polarity,
or 4th Order in like Polarity,
I can always identify the relative Polarity of one driver versus the other simply by listening to the loudspeaker playing music.
{I cannot always identify the crossover slopes' Order by listening.}
I have never tried listening to Sine Waves, or any other form of Test Signals,
when auditioning loudspeakers.
I think one may need to hear Transient dense program to hear this effect.
I hear it very distinctly with any percussion instrument,
and also with Piano music and any Orchestral that includes Pizzicato playing.
I have met people who can hear it.
Some are recording engineers.
If you don't hear it, consider yourself fortunate, as choosing or designing loudspeakers may be simpler for you than for those of us who do hear it.
*********************************************************
Hello tinitus,
good to read you here, and that you are continuing to trust your hearing !
Problem with these arguments is that it seems some people will not believe anything that they cannot experience themselves.
Human Nature has been this way since the earliest recorded evidence at least, and likely it will remain this way.
Myself, I do not intend to be a fool ... things do happen beyond my perception abilities.
I am posting here simply so that inquisitive people will not give up investigating the types of phenomena which may occur.
Last edited:
Would that imply that we also are supposed to hear a set of distributed tones in the same way the display on a spectrum analyzer looks like, which is absolutely stable, steady state?
If so, please listen to the two files of a slowly sweeping organ type of sound you can download here:
The-Ear-Is-Not-A-Frequency-Analyser.zip
(EDIT: Be careful with playback volume, files are almost 0dBFS level)
Anybody hear a systematic difference in timbre, also dynamically? Note it is not a subtle difference.
Yet both files are simple continuous multitone signals and look indistinguishable on the spectrum analyser, although their waveshapes radically differ in nature.
Klaus
Last edited:
You can have steep passive filters and still have 'transient perfect' response if you do it right. I explored one way to do it over at HTG.
HTGuide Forum - Duelund meets Dunlavy (aka Duelund meets transient perfect)
Indistinguishable on whose spectrum analyser? There is a difference even on my modest tools. Exactly what do you think this proves?
Still thinking what ears will do with some sort of signal??.... And the brain????
The point is just one " what will my brain do with the information it gets at time t?".
"If I listen to the same system and the same music in another room will my brain compute the same thing? Possibly not. Why?
Is that related to my physiological perception only or anything else?
Anyway, the best practical overall filter for me - in a multi-way system- is the 2nd order Bessel filter.
Cheers,
45
The point is just one " what will my brain do with the information it gets at time t?".
"If I listen to the same system and the same music in another room will my brain compute the same thing? Possibly not. Why?
Is that related to my physiological perception only or anything else?
Anyway, the best practical overall filter for me - in a multi-way system- is the 2nd order Bessel filter.
Cheers,
45
possibly, and it may actually be that in general
I like to be open to whatever works in any given situation(design)
it may not even one or the other
anyway, in real life, can you accurately really distinguish one from the other, and so precisely tell what you see in a response graph
well, now I have managed to get hold of a portable PC, so maybe I will learn how to do those measurements
thou, Im afraid they will only annoy me
I adjust my motor too
and have stopped using measurements, because it runs better without using that
well, measureing tools can be really nice, as a guide, to keep on track
but I dont trust them entirely, no
thou, if I were a pro in a workshop, I guess I would have to, yes
anyway, in real life, can you accurately really distinguish one from the other, and so precisely tell what you see in a response graph
No! In real life i just listen and decide (my brain decides...) what is more life-like!
Any response graph (especially, frequency response....) can be more misleading than anything else, for me!!!!
Cheers,
45
I'm not invalidating their designs - they've just chosen a different set of design compromises.
Also, good sounding does not necessarily mean accurate.
Last edited:
Hello David
If they sum to a flat frequency response what information is being discarded?
Acoustic phase, electrical phase both??
Still a 90 degree phase shift.
Rob
1 - transient information is being discarded - quite important I would think when we're talking about music . The "flat frequency response" you refer to is a steady state measurement.
2 - the total system output (sum of driver and crossover responses).
3 - the only phase shift in an ideal linear-phase speaker is a result of the bass response and the finite tweeter response.
The Duntech Soveriegn speakers, for example have a phase +/-25 deg from 80Hz to 18kHz. http://www.duntech.com.au/
My own Apollo speakers achieve a similar result.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
We could debate who hears what until the cows come home, and never come to a conclusion.
In my experience, I have found that speakers that use steep crossovers tend to have a hyper-detailed sound to them, whereas shallow-crossover designs, seem more relaxed and natural.
But thats just me, your mileage may vary.
It may be a direct consequence of linear-phase, or it could be an indirect consequence, like choice of driver materials, off-axis response, etc.
Last edited:
Ron,
What difference apart from slightly different spread distance for the harmonics do you see on your tools? I can guarantee that the spectrum is 100% clean and steady state in both cases because I wrote the software to generate these sweeps, any bouncing you might see comes from the chosen FFT-parameters.
What it proves? That the ear is not working as a mere spectrum analyser and is sensitive to actual waveform pretty much (which in your original post was said not to be the case), even with the most simple and benign steady state continuous signals I can think of.
Ah, btw, did you listen also (preferably before analysing)? If so, what was your listening impression?
Klaus
Last edited:
These are impressive curves but I assume a system with that much overlap betweens sections will have very poor vertical response. The Thiels suffer from that as well. You can pick your compromises, but I think with low order slopes you give up more than you gain.
David
David, I think the vertical response issues, though real, are exaggerated in practice.
Most people are sitting on their couch when listening to music, so the listening height is well defined and typically falls within a 1 foot window. At a listening distance of 9 feet or more, there is typically an acceptable +/- 2dB deviation in the response within that window.
The latest Thiels now use coaxial mid-tweeter drivers, but these have introduced a bunch of other compromises.
Most people are sitting on their couch when listening to music, so the listening height is well defined and typically falls within a 1 foot window. At a listening distance of 9 feet or more, there is typically an acceptable +/- 2dB deviation in the response within that window.
The latest Thiels now use coaxial mid-tweeter drivers, but these have introduced a bunch of other compromises.
Hello David
I am have a bit of trouble understanding what transient information is being discarded. Could you please clarify?
Here is a measurement of a clone system I built using 24db acoustic slopes. Speaker in my avatar. Looking at the measured response it meets the +/-25 deg from 80Hz to 18kHz requirement. Would you classify this as a linear phase design??
Rob
I am have a bit of trouble understanding what transient information is being discarded. Could you please clarify?
Here is a measurement of a clone system I built using 24db acoustic slopes. Speaker in my avatar. Looking at the measured response it meets the +/-25 deg from 80Hz to 18kHz requirement. Would you classify this as a linear phase design??
Rob
Attachments
Last edited:
It sounds like a modulated sweep, the second one is slightly different (dissonant), and looking at the waveform there is obvious "beating" going on. Certainly the envelope is different
IMO, beating is cheating, and a sweep is not "steady state", so that description of the signal is also disingenuous. If you think you have something, why not publish your research and assertions? Perhaps you have? Ich spreche Deutsch.The frequencies in the signals are different at the same time slots - the fundamental in the range 8-10 secs is ~41Hz, but in signal A the components are 41, 82, 123, 164, 205, 246 - a harmonic series
In signal 1, the frequencies in the same time period are 41, 84, 126, 167, 209, 257. This is not just a phase difference..... Because of this difference, I don't see anything here that interests me much from a psychoacoustic perspective.
If I analyse the whole waveform, there are very slight differences depending on which type of window and FFT size is used. But the ear doesn't average over 20+ seconds
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.