can phase delay show in rev in microseconds ?

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
Hello I mean REW not rev .I am not 100% sure if phase response and phase delay is diffrent display result. but in this university standford page phase delay display time and not angles. https://ccrma.stanford.edu/~jos/fp/Phase_Delay.html

I think the display of angles not very usefull. when it show in time is better readable to find fast mid range speaker that are good enough for ITD. I try group delay this give large delays on speaker over 1-3 ms at 500 hz because it seem measure more cycles. maybe ot get better results need measure at diffrent levels and look if the diffrence of the delay change much depend on volume. because for ITD should precise around 0.05ms and on stereo left and right speaker move diffrent ways, so delay between left and right speakers can change and give less stereo precision on slow speakers
 
Last edited:

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
Group delay should not change with volume since the response of the device should not change with volume, until it is driven outside its linear range.

Showing a phase delay figure would be easy, though it only shows the delay experienced by a single tone rather than seeing the delay of modulated content, which is what group delay shows. Calculating phase delay has a problem, however, because it is derived from the unwrapped phase and the result of unwrapping phase is dependent on where in the response the phase is considered to be within 1 cycle of the input. Measurement responses often go through many cycles of phase rotation, especially at low frequencies and whenever they pass through nulls, so identifying the true phase relationship is non-trivial. A different choice of where the response is considered to be within that single cycle gives a different unwrapped phase (shifting it by some multiple of 360 degrees) and so gives a different phase delay. Here are some examples, all from the same response, how does one decide which is the correct phase delay? The first is unwrapping from the start of the response.

43549


43550


43551


43552
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
the phase delay from 150 hz look real for me. it is only to compare speakers. how can i get such a phase delay display you post with 150 hz unwrapped display in REW ?. ITD get important from 150 hz upto 1 khz. and so 150 hz is a good starting point.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
You can't get phase delay, it is a feature I tried in a test build but is not in any release. I have not included it because of the phase unwrapping problem I explained above. In any case I don't see how it could be of any help for decisions related to ITD. If you want ITD to be preserved left and right drivers should simply have the same response. There is no such thing as a slow or a fast mid range driver, the terms are meaningless.
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
I think it is simular to slew rate in electronic circuits. woofer-mid driver that reach only 6 khz get maybe more diffrent time than 0.05 ms when left speaker need move 0.01 mm right speaker 0.1 mm per sample. on stereo systems left and right speakers move lots diffrent. so on slow slew rate their can happen additional phase diffrence because the speaker that need move more mm the result is more delay

ther can hear on big speaker that they sound as mono. this is an example video that show when range upto 450 hz is set mono. when play on small speakers hear more diffrence then play on large mid-woofer speaker also in compare to headphone.
 
Last edited:

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
You seem to be confusing different aspects of a driver's behaviour.

The driver's bandwidth determines the range of frequencies it can reproduce. The crossover is designed to make sure the driver is only fed signals whose frequency content lies within its bandwidth. At the upper end that is usually limited by cone breakup, at the lower end by the driver's resonant frequency, Xmax and the maximum sound pressure required. If a frequency is within the driver's bandwidth and the sound pressure required is not beyond the driver's power capability at that frequency it will be properly reproduced.

The amount the driver has to move to produce a given SPL depends on its effective area, the distance at which the SPL is measured and the square of the frequency of the sound. The formula can be found in this post (it can be re-arranged to derive the excursion for a given sound pressure). The bigger the effective area of the driver, the less it has to move to produce the desired sound pressure. Doubling the diameter reduces the excursion by a factor of 4, since area depends on diameter squared. Doubling the frequency also reduces the excursion by a factor of 4, since frequency is squared in the formula. For example, the movement required at 1 kHz is one sixty-fourth of the movement required at 125 Hz for the same sound pressure.

The driver excursion at the frequencies where ITD is relevant are usually very small.

ITD discrimination is, as I understand it, based on signal envelope, so group delay would be the relevant factor rather than phase delay.
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
I test group delay it have strange results on slow speaker and headphone the phase look ok. both headphone and speaker are record around 1.5 cm away on speaker in middle of woofer. the slow not wide sounding lp6 have best groupdelay but worsest phase. It have a DSP or is it possible that there can fake with DSP that the group delay look good but phase and sound not good. ?. the spikes are in because i use EQ for measure to correct speakers. without EQ the spikes are gone and excess GD stay near same
eris high volume no eq phase.jpg
eris 3.5 low volume GD.jpg eris 3.5 high volume GD.jpg
eris 3.5 high volume noeq GD.jpg
stay near same. see screenshots on eris

rp th030 headphone GD.jpg
rp th030 headphone phase .jpg
kali lp6 1.5 cm GD.jpg
kali lp6 1.5 cm phase.jpg
 

Attachments

  • eris high volume no eq phase .jpg
    eris high volume no eq phase .jpg
    237.3 KB · Views: 15

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
I test group delay it have strange results on slow speaker and headphone the phase look ok. both headphone and speaker are record around 1.5 cm away on speaker in middle of woofer. the slow not wide sounding lp6 have best groupdelay but worsest phase.
By "lp6" do you mean the Kali Audio LP-6? Why do you keep calling it "slow"? It has a frequency response that goes well beyond 20 kHz per the published specs, it is 'faster' than your ears :)

If you are seeking some measurement that will explain why a loudspeaker doesn't sound like a headphone I think you are wasting your time. They operate in totally different environments, why should there be anything comparable about their measurements? Why should such different listening experiences be perceived as being at all the same?

It have a DSP or is it possible that there can fake with DSP that the group delay look good but phase and sound not good?
No, that's a nonsensical statement. Group delay is simply the slope of the phase negated.
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
yes i mean kali LP6.on step response see that it is slower. mean that only the woofer is too slow for ITD. I have no other logical explain for this effect, that large woofer mid speakers often sound as a dereverb and let the room and stereo width sound smaller and muddy. I of course measure the speaker and correct them, but stereo width is much smaller this happen also when i do a lowpass of 1.4 khz -24 db that stereo width sound reduce on LP6. so the tweeter have no influence to that.

when have double track guitar 2 diffrent versions 1 left 1 right should sound same as in headphone that both can hear full left and full right. on lp6 it sound as it come from middle and not so loud(compare with the highhat) and much smaller. thats wrong . part 2 b is the lp6. i record the speaker.

here i have diffrent audio records. I zip it because forums do not support mp3 and large files so you can hear. the lp6 sound really worse for me. I have in past a tannoy nfm 8 mk2 a 8 inch coaxial speaker this was even more worse in stereo width. then i notice that large mid woofer 2 are the problem. on 3 way system i hear such problems not or on small 2 way speakers. so my question to solve is how can see iin measurement if a larger mid woof speaker is maybe fast enough for optimal stereo width

you can hear the audio examples with headphones and compare the stereo width with the original. I hear lots diffrence in the lp6 record in stereo width
 

Attachments

  • Part 1 b(original).zip
    575.4 KB · Views: 11
  • Part 2 b.zip
    574 KB · Views: 8
  • eris nt.zip
    562.9 KB · Views: 7
Last edited:

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
It doesn't make sense to try and compare the perception of listening on headphones to listening on speakers. With headphones the left ear gets only the left channel, the right ear the right. With speakers both ears hear both channels. There are so many differences in what the ears and brain are presented with in both cases it is amazing that the brain is adaptable enough to make them sound similar to us.

Your premise that the differences you perceive are due to some characteristic of the driver seems flimsy. In particular what comes across as an obsession with 'speed', presumably being a term you are using the describe the bandwidth of the driver, does not appear to have any logical basis. The crossover determines which parts of the input signal go to which driver, and the overall frequency response shows how well that does or doesn't work.

The step response doesn't tell you much about the bandwidth of the system, its appearance tends to be dominated by the affects of driver time alignment. If the drivers in a speaker are not time aligned (often the case) the contributions of woofer and tweeter are separated in time slightly. The effect on the appearance is a lot greater than the subjective impact. The easiest way to tackle that is to try something like Dirac, which will bring the response largely into alignment as part of its processing.

Surely a more practical way of dealing with your image width concerns would be to listen to different speakers until you find one whose presentation you like?
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
sure on the speakers the diffrence is much more hear. but also on headphonse the diffrence can hear alot when can hear ITD. please hear the examples and compare. did you not hear a diffrent in stereo width ?. you can also use my orignal and play on diffrent speakers. if you have some speaker with diffrent size maybe you come to same result that the speakers with big woofer mid have less stereo width,

because ITD is hear from 150 hz upto 1.5 khz and humans are able to hear at least 50 microseconds in phase diffrence in the frequency the logical question for me is. How can a bass-mid woffer that can only reach frequency upto 6 khz produce the midrange good in phase at least 0.05 ms. 6 khz is a period time of 0.16 ms

i think when a speaker can produce correct phase of 0.05 ms this speaker is also able to priduce frequency with period time of 0.5 ms. and period to frequency calculator say 20 khz is period time of 0.05 ms

http://www.loudspeakerdatabase.com/Dayton/DC160S#

do you think such a speaker can produce correct phase with less diffrence as 0.05 ms from 150 hz - 1.5 khz between left and right speakers ?

here can see many speakers with FR and type


in the view click on blue dots below chart can change . if no ITD can hear this speaker ist good enough for 3 khz crossover. When you can hear ITD and you want have correct stereo width choose a mid woofer that go much higher. maybe 12 khz

problem is only. on non self made speakers there can not see which FR the woofer do, because of the crossover
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
because ITD is hear from 150 hz upto 1.5 khz and humans are able to hear at least 50 microseconds in phase diffrence in the frequency the logical question for me is. How can a bass-mid woffer that can only reach frequency upto 6 khz produce the midrange good in phase at least 0.05 ms. 6 khz is a period time of 0.16 ms

i think when a speaker can produce correct phase of 0.05 ms this speaker is also able to priduce frequency with period time of 0.5 ms. and period to frequency calculator say 20 khz is period time of 0.05 ms
You are mixing up several things there, and coming to an illogical conclusion. The ability to reproduce signal timing has nothing to do with the bandwidth of the driver.

Here are two 1 kHz sine waves with a 50 us time difference between them.

43670


If I want to reproduce those signals and that time difference, all I need is a pair of speakers that can reproduce 1 kHz. The time difference between the two channels doesn't make them any easier or harder to reproduce and doesn't mean the speakers need to be any more or any less capable. If they can reproduce 1 kHz, they can reproduce 1 kHz signals with any time difference you like. 1 us, 1 ms, 1 s, 1 minute, it doesn't matter. Send them signals with the delay you want and that's what you'll get.

All signals can be decomposed into the frequencies that make them up, that is what the FFT does. To preserve the relative timing of signals in the 150 Hz to 1.5 kHz range, the system just needs to be able to reproduce that range. It doesn't even have to do it with one driver, it could spread the load among several drivers each covering part of the range as long as the crossovers were designed so that the outputs sum correctly. As an example, I generated a synthetic 2-way speaker with a woofer that covers 50 Hz to 2 kHz and a tweeter that covers 2 kHz to 20 kHz, and compared its step response to the step response of a synthetic single driver that covers 50 Hz to 20 kHz. Here are the two step responses, which is which? Where is the 'slow' woofer?

43673



Again, if you don't like the way your LP-6 sound why don't you just listen to other speakers and find something you do like? Why waste time inventing nonsensical 'explanations' for your perceptions?
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
but between kali and other speaker the step response is much more difffrent. the best i have test is the JBL 104 BT. I think to see if a speaker is good the time form 40% level to reach max level and the time to decay to 40% level show the speed/precision of a speaker in mid range. I correct the 104 with EQ to be in FR simular to LP6. wy look on lp6 rise and fall time much slower ?. position of speaker and microphone is not exact same but very simular. what do you think wy lp6 and the 104 is so much diffrent ?
jbl measure.jpg
lp6 measure.jpg


larger speakers are better in bass so if a large speaker bring same stereo width as a small speaker i prefer of course the large speaker. there should be a measurement on that can see if the speakers have good or bad stereo width.
if the bad stereo width does not happen because of slow mid range speaker, wy you think there is diffrence in stereo width on the example songs and when hear them on the speaker ?. the diffrence is very big. I have also done blind tests and i hear this.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
I think to see if a speaker is good the time form 40% level to reach max level and the time to decay to 40% level show the speed/precision of a speaker in mid range.
It doesn't. It is contradictory to talk about 'speed' and 'mid range'. Fast rise comes from extended high frequency response. Fast decay comes from poor low frequency response.

The main difference between a large 2-way and a 3-way or small 2-way is not the bandwidth of the driver reproducing the mid range, it is that the driver reproducing the mid range does not also have to reproduce low frequencies. Perhaps your answer lies there. If you use headphones as a reference you will never find an answer, however. It is impossible for a speaker to be heard by only one ear, comparing stereo impressions for headphones and speakers is completely meaningless.
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
and what do you think, wy the step response rise and fall time of the JBL and kali is so much diffrent ?. borth have crossover around 1.6 khz you can see also FR is very simular. see the diffrence between the 2 arrows i mean. when i use no eq on the JBL does not change much on the step response. maybe the problem for stereo is when a speaker do overshoot.and a slow speaker that can only reach 6 khz do more overshoot undershoot because it is not so precise. the kali show much overshoot in step response at beginning.
I hear that the jbl or other small speakers sound most near to headphone. on large woofers i hear less stereo width.
step response.jpg


here is the JBL 104 BT without EQ. much less bass, but step response stay near same.
response noeq.jpg
 
Last edited:

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
Different crossover types, different physical arrangement, different drivers, different enclosures. You should give up on your obsession with trying to divine something from an entirely inappropriate metric.
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
I have edit my previous post and add the overshoot undershoot problem. that a heavy and large speaker do overshoot undershoot seem logical. and a speaker that do undershoot/overshoot seem maybe not able to produce a waveform correct for 0.05 ms.

I have a 2 channel USB osciloscope. so i can overlay original signal with recordet signal 1 cm away. there can see much more diffrence on the recordet waveform (bursts). the delay of peak between original differ depend on frequency that play the tone and volume. only problem how can this see on existing speaker measure software.
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
there can see much diffrent group delay. on the speaker with good stereo width (JBL 104 BT) this is much less. but when do excess group delay then is not much diffrence see. I measure both speakers 1 cm away and i use 150 HZ LK24 higpass and 1.15 khz LK24 Lowpass to reduce FR influence for both speakers.

Wy can the excess group delay on the LP6 even if it is more worse in group delay look nice ?
i attach also group delay measure of headphone. measure distance
lp6 GD.jpg
104 BT.jpg
on headphone is too 1 cm
heaphone GD.jpg
 
Last edited:

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
137
Group delay should not change with volume since the response of the device should not change with volume, until it is driven outside its linear range.

Showing a phase delay figure would be easy, though it only shows the delay experienced by a single tone rather than seeing the delay of modulated content, which is what group delay shows. Calculating phase delay has a problem, however, because it is derived from the unwrapped phase and the result of unwrapping phase is dependent on where in the response the phase is considered to be within 1 cycle of the input. Measurement responses often go through many cycles of phase rotation, especially at low frequencies and whenever they pass through nulls, so identifying the true phase relationship is non-trivial. A different choice of where the response is considered to be within that single cycle gives a different unwrapped phase (shifting it by some multiple of 360 degrees) and so gives a different phase delay. Here are some examples, all from the same response, how does one decide which is the correct phase delay? The first is unwrapping from the start of the response.

View attachment 43549

View attachment 43550

View attachment 43551

View attachment 43552

Hello ,
I get in mind maybe the delay can output in microseconds when the cursor is on a frequency placed in the phase view the phase excess phase ? . maybe can in the phase excess output also the value in microseconds stand ?. so text is then Excess Phase -159 deg / xxxxx µsec .

excess phase.jpg
 
Top Bottom