step response is a important part to show speed of speaker that is good enough for ITD. See measures

[bennybbbx]

I measure a headphone with Low Pass Filter 6 or 12 db at 1.5 khz to simulate a crossover. I do the test also with Phase linear filter, but it change not much in step response so i do not upload the results. there can see the headphone with the simulate crossover is much more faster as the speakers. The headphone is measure 1.5 cm distance and the speakers too. there can see the LP6 is horrrible slow in compare to headphone. The JBL 104 BT is a coaxial system so it show always the tweeter in front. the JBL give the best stereo width of all speakers i test. it have fastest step response. The JBL i have also measure with speaker correction i use in my room. this is named as EQ measuret. you can see, the step response did show only small chamges when i measure the JBL with EQ or without EQ. the LP6 give very few stereo width because woofer/mid is so slow is my conclusion i get after test of 5 speakers with diffrent mid/woofer sizes

headphone fullrange no smoothing. other measures have for the frequency 1/2 octave smoothing. choose of smoothing does not change anything in step impulse display. it is only to see how the headphone look unsmoothed. half octave smoothing i think best view

Headphone LP 1.5 khz -6 db

Headphone LP 1.5 khz -12 db

headphone LP 2khz -12 db

Kali LP6

JBL 104 BT

. maybe other that can hear ITD can verify with tests of their own speakers (NO THEORY) if they come to same conclusion.

Conclusion is the LP Filter of crossover give aorund 0,4 ms physical step delay. but the speakers have much longer delay to reach at least 20% level again. and when compare speakers the speaker that have the nearest time to the headphone impulse with the 1.5 khz -12 db crossvover impulse response is the best. speakers with crossover freq 2 khz or more have pysical shorter step delay. i only upload 1 measure of 2 khz -12 db to show this. so there need much improve for good speakers for people that can hear interaural time delay



headphone 400 hz lowpass. you can see so slow is the LP6

the result show that there can much improve for speakers.

 

[ppataki]

Hi @bennybbbx I also came to the same conclusion when measuring my DIY subs: HPF and LPF filters do cause a degradation in the step and impulse response curves. Actually I was using the Wavelet diagram to showcase the same (cannot find that thread now but I posted it a while ago)

 

[bennybbbx]

Hi @ppataki

for subwoofers low frequency crossover the delay is a more important thing.. thats maybe the reason because 3 way system use often high crossover of 400-800 hz. I do a measure with 12 db lowpass at 70 hz . but it show no usefull result with headphone. to look if the speaker is a speedproblem you need the step repsonse time that get from the speaker subtract from a headphone(or a excellent speaker) measure after a lowpass. or just use my example measures for subtract to see what can reach. then you see how much slower a speaker is.you can see the speakers are much slower and between 2 khz and 1.5 khz crossover is not much dffrence. the much diffrence can see on speaker impulse

I do also measure headphone and 400 hz LP 12 db. i have add it in 1. Post now too .this have simular step response as the LP6.

wavelet diagram is better but need a easy to display thing all speaker measures soft have and the klippel software can do. so the step response is usefull to have in speaker tests for 2 way speaker. with impulse response it is not so easy to see the diffrences.

the measurement i do with my additonal channels. I have a focusrite 8i6 USB interface.

I use output 5 on rew for measure output. In focusrite mixer i choose for channel 5/6 loopback 1/2 . this is digital loopback. on DAW i use loopback 1/2 as input and mix it to master out with 100% .In master out is the plugin to simulate the filters. master out go to speakers or headphones

 

[thewas]

I measure a headphone with Low Pass Filter 6 or 12 db at 1.5 khz to simulate a crossover. I do the test also with Phase linear filter, but it change not much in step response so i do not upload the results. there can see the headphone with the simulate crossover is much more faster as the speakers. The headphone is measure 1.5 cm distance and the speakers too. there can see the LP6 is horrrible slow in compare to headphone. The JBL 104 BT is a coaxial system so it show always the tweeter in front. the JBL give the best stereo width of all speakers i test. it have fastest step response. The JBL i have also measure with speaker correction i use in my room. this is named as EQ measuret. you can see, the step response did show only small chamges when i measure the JBL with EQ or without EQ. the LP6 give very few stereo width because woofer/mid is so slow is my conclusion i get after test of 5 speakers with diffrent mid/woofer sizes

headphone fullrange no smoothing. other measures have for the frequency 1/2 octave smoothing. choose of smoothing does not change anything in step impulse display. it is only to see how the headphone look unsmoothed. half octave smoothing i think best view
View attachment 168985


Headphone LP 1.5 khz -6 db
View attachment 168989


Headphone LP 1.5 khz -12 db
View attachment 168988

headphone LP 2khz -12 db

View attachment 168984

Kali LP6


View attachment 168991

JBL 104 BT

View attachment 168987

View attachment 168986. maybe other that can hear ITD can verify with tests of their own speakers (NO THEORY) if they come to same conclusion.

Conclusion is the LP Filter of crossover give aorund 0,4 ms physical step delay. but the speakers have much longer delay to reach at least 20% level again. and when compare speakers the speaker that have the nearest time to the headphone impulse with the 1.5 khz -12 db crossvover impulse response is the best. speakers with crossover freq 2 khz or more have pysical shorter step delay. i only upload 1 measure of 2 khz -12 db to show this. so there need much improve for good speakers for people that can hear interaural time delay



headphone 400 hz lowpass. you can see so slow is the LP6

View attachment 169005

the result show that there can much improve for speakers.

Is it only me having a very hard time trying to understand the text?

Since it seems you are a non-native English speaker (same like me), I would recommend you formulating short and precise sentences in your mother language and translating them to English using Google or even better DeepL.com translators which works very good for German=>English. Also its better if for every plot you write the the corresponding text directly before or after it and not first all the text together followed by all plots.

 

[ctrl]

I measure a headphone with Low Pass Filter 6 or 12 db at 1.5 khz to simulate a crossover....there can see the headphone with the simulate crossover is much more faster as the speakers. The headphone is measure 1.5 cm distance and the speakers too. there can see the LP6 is horrrible slow in compare to headphone.

I can't gauge it, are your statements meant as satire or is this an attempt at trolling?

After many people in the neighboring thread have pointed out to you, over weeks, the problems with the evaluation of the step responses over and over again, you post again (I can't say it any other way and apologize in advance for it) such crap.

Even after months, you still haven't looked at the connection between bandwidth and rise time.
Only if the frequency responses of the measured loudspeakers or headphones are congruent, you can compare the step responses the way you do.

Normally one assumes that drivers behave (almost) as a minimum phase system in their transmission range. This means that with identical frequency responses of two drivers, congruent step responses are obtained.

In other words...
If the frequency responses of two drivers (e.g. headphones and woofer) are identical (as congruent) over a bandpass, e.g. 400-2000Hz, you could then compare the step responses directly.
Any deviations in the step responses would then be a non-minimum phase behavior (from one or the other or both), which is usually represented as excess phase in the phase frequency diagram.

Your measurements must also not include room resonances, so you can only compare near-field measurements (<1cm distance) or gated measurements at frequencies about >1kHz (depends on our room and test setup).

At least read up a little on how to use REW. Then you can show two frequency responses at the same time as overlays and we can more easily see if they are identical.
If this is the case, you can show us the step responses as overlays and then we can see if there are major differences in the rise time.

... and please use the thumbnail function of the forum software to display your measurements.
Half a dozen huge pictures in a row, without any text and explanations are overkill.

 

[bennybbbx]

I can't gauge it, are your statements meant as satire or is this an attempt at trolling?

After many people in the neighboring thread have pointed out to you, over weeks, the problems with the evaluation of the step responses over and over again, you post again (I can't say it any other way and apologize in advance for it) such crap.

Even after months, you still haven't looked at the connection between bandwidth and rise time.
Only if the frequency responses of the measured loudspeakers or headphones are congruent, you can compare the step responses the way you do.

Normally one assumes that drivers behave (almost) as a minimum phase system in their transmission range. This means that with identical frequency responses of two drivers, congruent step responses are obtained.

In other words...
If the frequency responses of two drivers (e.g. headphones and woofer) are identical (as congruent) over a bandpass, e.g. 400-2000Hz, you could then compare the step responses directly.
Any deviations in the step responses would then be a non-minimum phase behavior (from one or the other or both), which is usually represented as excess phase in the phase frequency diagram.

Your measurements must also not include room resonances, so you can only compare near-field measurements (<1cm distance) or gated measurements at frequencies about >1kHz (depends on our room and test setup).

At least read up a little on how to use REW. Then you can show two frequency responses at the same time as overlays and we can more easily see if they are identical.
If this is the case, you can show us the step responses as overlays and then we can see if there are major differences in the rise time.

... and please use the thumbnail function of the forum software to display your measurements.
Half a dozen huge pictures in a row, without any text and explanations are overkill.

are the images not self explain ?. they show that on a fast low/mid driver (the headphone) at a crossover frequency of 1.5 khz step response have a a time of 0.4 ms. the kali have time of 1.8 ms. so the kali need 4.5 * longer. i told before the changes that happen due to LP filter is so minimal in compare. but we calc it. think about a LP filter have 50% slower step response and this is 0.6 ms and the speaker still 1.8 ms. then the speaker is still 3* slower than the headphone.

when you want more precise measures you see the crossover frequency of the speaker. then measure the headphone with this crossover frequency . I or something other can do headphone measurements of crossover from 1.3 khz upto 3 khz in 100 hz steps and for diffrent orders 6 12 18 24 db. this is of course some work. but you can then subtract the measured value of the fast headphone and get a result value that show how much the speaker is slower as the headphone

sure this is not a 100% exact measure maybe it change a little if BW or LK filter or is only usefull for 2 way speakers. but i ask aften. which measure value show better if a speaker have good or bad stereo width ? . the JBL have 1.0 ms and this is for me the best stereo width. maybe there are faster speakers.

 

[ernestcarl]

bennybbbx is convinced that "good stereo width" (soundstage, imaging, and localization) outcome/perception is divined by analyzing step response graphs. Some filters also negatively "slow" the step response which in turn affects this so-called stereo width. Various headphones and the small JBL 104 coax speakers he uses are much "faster" than his 2-way Kali LP-6 monitors.

It is just a very folk misunderstanding of how sound and psychoacoustics work.

 

[j_j]

I think that when an understanding of the actual relationship between bandwidth and impulse response shows up, I'll consider commenting. Ditto a good understanding of linear superposition

 

[bennybbbx]

Now i write the step response time the system need from 20% to 100% and back to 20% Level that can see in the measure pictures. Maybe then it is more clear how can see if a speaker is slow or not. if somebody want have the measure dat let me know. i can not upload here it is 8 mb zip.

For information 1000 µ sec period time is a frequency of 1 khz.

280 µsec time the 1.5 khz -6 db Lowpass before headphone(headphone have much more Low bass as the Kali)
366 µsec time the 1.5 khz -12 db Lowpass before headphone
303 µsec time the 2 khz -12 db Lowpass before headphone

1448 µsec Kali LP6

900 µsec the JBL with EQ that boost bass by 12 db have more bass as Kali and some other frequency bands around 3-6 db for FR correction
800 µsec the JBL without EQ have lower bass as the Kali

you see the step response of the Kali is 1448/366 = 390% slower as the Headphone with LP Filter.

when want compare the diffrence in step response time between much EQ on the JBL and no EQ on JBL it is 900/800= 12%

when want compare the diffrence between 6 db to 12 db 1.5 khz LP filter (we know Kali have 1.5 khz crossover).
when measure the woofer at distance 1.5 cm then can see the frequency fit to a 12 db crossover LP filter

ok maybe it change a little if BW or LK or BL Filter Q then add another 8 % of unknown influence

So if people say my conclusion is wrong, then they must also say which LP filter can increase the step response time
from lets say 400 µsec to 1400 µsec at 1.5 khz ????????

It should also clear that a speaker + filter that need 1400 µsec for a step response 20% to 100% and back to 20% is much less able to produce
the phase correct for 40 µsec that at least need for good ITD hearing.
in wikipedia about step response stand this

What does the step response tell us?




Formally, knowing the step response of a dynamical system gives information on the stability of such a system, and on its ability to reach one stationary state when starting from another.

stability can not test on speakers . but we need to know how fast a system can reach levels and decay. of course filters do too delay the step response together with the speed of the speaker system. The speaker system, is as can see on my measures the most important slowdown part. i think it is strange that the impulse response view output not good raise and fall time curves. with step response can see better the diffrences

 

[bennybbbx]

This means that with identical frequency responses of two drivers, congruent step responses are obtained.

If the frequency responses of two drivers (e.g. headphones and woofer) are identical (as congruent) over a bandpass, e.g. 400-2000Hz, you could then compare the step responses directly.
Any deviations in the step responses would then be a non-minimum phase behavior (from one or the other or both), which is usually represented as excess phase in the phase frequency diagram.

I try my best to get a simular step response as the kali with lots EQ correction (-36 db max and which begin at around 130 hz so i can reach that kali step response with headphone ) on not linear phase. FR is diffrent to Kali. then i use for the result same EQ and FR Linear phase and look how the step response look very diffrent but the FR is the same as the non linear phase. with the button minimal phase you can switch between. conclusion for me is, the step response is usefull, maybe kali do some worse phase shifts in low freq EQ correction. room reverb is no
problem when measure at 1.5 cm. see which heavy EQ correction i need to reach such a step response

minimal Phase EQ

Linear phase EQ

this measurures show, it is not only the FR it is also the phase shift of EQ which make the step response worse. anyway the kali sound not wide. so wy not show a step response for speakers.

EDIT: here is a overlay compare between the FR of kali and the much EQ headphone. when use less smothing can see kali do from 1.5 khz to 2 khz -20 db. thats a very high order filter. and after 2 khz it stay on this level. maybe the phase shift of this sharp filter make it sound so bad. good that i notice it, because then it make no sense to try a 6.5 inch wide band speaker in the Kali

 

[tomtoo]

bennybbbx is convinced that "good stereo width" (soundstage, imaging, and localization) outcome/perception is divined by analyzing step response graphs. Some filters also negatively "slow" the step response which in turn affects this so-called stereo width. Various headphones and the small JBL 104 coax speakers he uses are much "faster" than his 2-way Kali LP-6 monitors.

It is just a very folk misunderstanding of how sound and psychoacoustics work.

Ok, than explain. I realy dont like if people tell, i know and he is stupid.
Explain man, if you not see it the same way. Explain where he is wrong?
What he likes to tell is, that a speaker needs a time to come to full swing power. The lower membrane mass, magnetic power of the motor, and lower airmass the membrane has to move, the faster this goes. And the hearing of us is very sensitive for direction determination to the transient response, and this has some indluence on percived stereowith. So if you say thats BS than please take the time to explain. I dont know if this is right or wrong, BUT explain!

 

[NTK]

Ok, than explain. I realy dont like if people tell, i know and he is stupid.
Explain man, if you not see it the same way. Explain where he is wrong?
What he likes to tell is, that a speaker needs a time to come to full swing power. The lower membrane mass, magnetic power of the motor, and lower airmass the membrane has to move, the faster this goes. And the hearing of us is very sensitive for direction determination to the transient response, and this has some indluence on percived stereowith. So if you say thats BS than please take the time to explain. I dont know if this is right or wrong, BUT explain!

The most obvious error is that ITD is interaural time difference. It only happens when there is a difference in the timing of the left speaker and the right speaker. If the two speakers are the same and have the same "slowness", there is no interaural time difference -- they are equally "slow" and the sound waves originated from both speakers arrive to both ears at the same time (assuming listening position is equal distance to the 2 speakers).

It has been pointed out by JJ to OP in this post.

Step Response: Does It Really Matter?

yes i have listen. i want wait what others hear. others do not report what they hear. now i write. I hear in the AM tone huge wobbling(i think it wobbles around 5 hz, but i am not sure. can also be 10 hz) with headphone and my jbl same strong. the fm.wav is more clean but it wobbles too clear...

www.audiosciencereview.com

 

[bravomail]

Hi @bennybbbx I also came to the same conclusion when measuring my DIY subs: HPF and LPF filters do cause a degradation in the step and impulse response curves. Actually I was using the Wavelet diagram to showcase the same (cannot find that thread now but I posted it a while ago)

Yes. Step impluse response will show the "problems" for crossovers in speakers. Cue in "active crossovers" which will solve such issues! expensive, though...

 

[tomtoo]

The most obvious error is that ITD is interaural time difference. It only happens when there is a difference in the timing of the left speaker and the right speaker. If the two speakers are the same and have the same "slowness", there is no interaural time difference -- they are equally "slow" and the sound waves originated from both speakers arrive to both ears at the same time (assuming listening position is equal distance to the 2 speakers).

It has been pointed out by JJ to OP in this post.

Step Response: Does It Really Matter?

yes i have listen. i want wait what others hear. others do not report what they hear. now i write. I hear in the AM tone huge wobbling(i think it wobbles around 5 hz, but i am not sure. can also be 10 hz) with headphone and my jbl same strong. the fm.wav is more clean but it wobbles too clear...

www.audiosciencereview.com

Ok, but lets see that stereo speakers get not the same signal, its different for l,r channel. And the 'slower' the speakers are the more this timediffernces lead to a smeared stereo impression?

 

[ernestcarl]

Ok, than explain. I realy dont like if people tell, i know and he is stupid.
Explain man, if you not see it the same way. Explain where he is wrong?

It is wrong to draw such over-arching "stereo width" conclusions by just looking at those graphs. Headphones and speakers, for instance, are very, very different beasts -- there is no crossfeed in the former. The headphones would sound "wider" simply by that fact alone... in the other thread it was also mentioned before that other parameters significantly mattered more like directivity characteristic of the speaker system.

 

[tomtoo]

It is wrong to draw such over-arching "stereo width" conclusions by just looking at those graphs. Headphones and speakers, for instance, are very, very different beasts -- there is no crossfeed in the former. The headphones would sound "wider" simply by that fact alone... in the other thread it was also mentioned before that other parameters significantly mattered more like directivity characteristic of the speaker system.

Clear, you cant compare the stereo wide of a headphone with a speaker. We even do crossfeed to correct this. But at the end i not see a answer if a 'faster' speaker could give the impression of better stereowide or not?

 

[ernestcarl]

Clear, you cant compare the stereo wide of a headphone with a speaker. We even do crossfeed to correct this. But at the end i not see a answer if a 'faster' speaker could give the impression of better stereowide or not?

I think it is desirable to have a more linear time response, for sure. But I highly doubt one would automatically get the impression of better "stereo width" perception by just that one property. I could, for instance, physically position speakers closer or farther apart to change the perceived stereo separation.

 

[NTK]

Ok, but lets see that stereo speakers get not the same signal, its different for l,r channel. And the 'slower' the speakers are the more this timediffernces lead to a smeared stereo impression?

Don't see why the speaker "speed" matters. I am "borrowing" these figures from one of Dr. David Griesinger's papers**.

http://www.davidgriesinger.com/threedpm.pdf

Let's ignore the center channel mic and speaker in this discussion and the listener is equidistance from the left and right speakers. The left speaker will be reproducing the violin 6 ms before the right one. When the left and right speakers are the same (and has the same delay), the sound picked up by the left mic will be reproduced by the left speaker exactly the same 6 ms before the right speaker reproducing the right mic signal. The listener will hear the correct ITD.

** Griesinger used these figures to show why time delay panning only works well for listening positions on the line between left and right speakers. For home theatre applications intended for multiple listeners, only amplitude panning works.

 

[bennybbbx]

It is wrong to draw such over-arching "stereo width" conclusions by just looking at those graphs. Headphones and speakers, for instance, are very, very different beasts -- there is no crossfeed in the former. The headphones would sound "wider" simply by that fact alone... in the other thread it was also mentioned before that other parameters significantly mattered more like directivity characteristic of the speaker system.

before you judge about me, maybe you can test a speaker with a fast stepresponse as the JBL 104. maybe you can post your step responses without EQ correction
I was really surprised that the jbl sound wider as all headphones i have. it is easy to compare and much wider as all the headphones i have. i hear in headphones not much wide diffrences. I have some headphones. the headphones of course sound much wider as the kali and also a little wider as the iloud mtm so i sell it and presonus with tweeter i have modified that it can work as wide band speaker.

maybe it is important to use fast filter. JBL have only 6 db filter. you see in my example 6db have fastest attack time of all filters i post. I do tests with ribbon tweeter and i notice that crossover of 24 db sound really bad for me. and it seem the kali lp6 use 48 db filter.

I have a demo of split EQ since some time. this can simulate a 48 db high shelf filter. because the level drop 20 db and then stay at same they seem use a high shelf and not a Low pass

i do measure and i notice step response get very worse. slow rise time. but it is still not so slow as the Kali. the step response have prering and this EQ have not zero latency. so it is some kind of linear phase filter. and BTW: if the filter or the speaker make it slow doesnt matter because it sound worse. much phase shifting in filters do damage ITD delays between speakers. there happen never same frequency and level on both speakers. I have no other demo vst that support 48 db filter to test with currrently. but i look for one to test later how the step response look with a 48 db non linear phase high shelf filter

does for others that can hear ITD 48 db filters sound good ?. I am really surprise that 48 db filters give so slow step response.

 

[bennybbbx]

Don't see why the speaker "speed" matters. I am "borrowing" these figures from one of Dr. David Griesinger's papers**.

http://www.davidgriesinger.com/threedpm.pdf

View attachment 169415

Let's ignore the center channel mic and speaker in this discussion and the listener is equidistance from the left and right speakers. The left speaker will be reproducing the violin 6 ms before the right one. When the left and right speakers are the same (and has the same delay), the sound picked up by the left mic will be reproduced by the left speaker exactly the same 6 ms before the right speaker reproducing the right mic signal. The listener will hear the correct ITD.

** Griesinger used these figures to show why time delay panning only works well for listening positions on the line between left and right speakers. For home theatre applications intended for multiple listeners, only amplitude panning works.

did you hear that diffrent speakers sound at nearfield 80 cm left to right distance diffrent in width and depth of field ?. this is also written in tests. the JBL 104 are also test with good stereo width. the kali some find good, some bad. when the reason for good stereo wide is not a fast step response, what reason do you think it is and with which measure value can see a good stereo width. what do you think ?

EDIT: read also my above post. and additional . the headphones are more near in width as the speakers with 80 cm left to right . and i can hear if a room is big or small in speakers better. it sound as sit in the room i hear also reverb come from back, but i can not locate virtual surround(that voices come from back). but maybe with a speaker with faster step response can. with kali all sound as i hear thru a door with very few stereo width at 80 cm. maybe when put them 2 meter from left to right it es better. but the kali is sell as near field monitor