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

[bennybbbx]

For the third time. Both of your microphones are omnidirectional. This is stated by the manufacturers of both microphones themselves.

Neither of them are ‘unidirectional’, as this would be a literal laser beam.



Incorrect. Here is an example. I measured a single Faital Pro 3FE22 driver resting on it's box on my desk, with an Earthworks M30 microphone at 10 centimetres, using an 87.4 second stimulus with the AFMG EASERA TDS module. No post-processing windowing is applied.

Due to the way Time Delay Spectrometry works, we can consider this to be about as a reflection-free measurement as possible in the circumstances.

I then duplicated the trace, and edited the data directly to set the magnitude from 30 Hz to 200 Hz to a flat value at 91 dB. Markers are added to the traces to show the level increase over the original measurement.
View attachment 175440

This modification is linear phase, as you can see here in the phase trace of both data sets. The same cursors are visible, confirming identical phase at the adjusted frequencies:
View attachment 175441

Here are the two impulse responses, overlaid
View attachment 175443

You can barely see the shift in the blue modified data, so here it is zoomed in. The later part of the IR deviates more, as you'd expect from increased energy.
View attachment 175442

And here we have the two step responses. Again, blue trace is the artificially modified magnitude response with much more energy in the sub region, but no change in phase:
View attachment 175444

your step response view does not show good the range from 0-1 ms. only this range is important for mid speed. i have upload rew another measure file, also do with a diffrent position. step response stay same. try yourself. linear phase filters do prering, this can also influence the step response. just try the test with a subwoofer and without subwoofer and do not change EQ and show the step response range from 0- 4 ms

 

[dasdoing]

Conclusion: Step Response shape (or more correctly, excess phase contribution) has some impact on imaging but it is not the key parameter.

It's not so much about perfect step-response or excess phase elimination. the important part is phase coherence between the speakers (in room). the more they match, the more the speakers disapear and deeper/3d the soundstage becomes

 

[Wesayso]

Benny, I can't wait until you discover the wavelet tab within REW. If you want a visual guidance of timing, the wavelet doesn't disappoint!

Here's an example of an ideal IR band-passed model:

The frequency response where it's easy to see the band passed output:

The STEP and IR that belongs to the above graph:

That's a pretty ideal result, right? So what does the Wavelet look like for a result like this:

I included the settings, so you can compare your measurement with the exact same settings and really see the timing you get at the listening spot. Don't be too disappointed by the outcome, it might throw your whole theory down the drain...

 

[bennybbbx]

Benny, I can't wait until you discover the wavelet tab within REW. If you want a visual guidance of timing, the wavelet doesn't disappoint!

Here's an example of an ideal IR band-passed model:

The frequency response where it's easy to see the band passed output:
View attachment 175553

The STEP and IR that belongs to the above graph:
View attachment 175554
That's a pretty ideal result, right? So what does the Wavelet look like for a result like this:
View attachment 175555
I included the settings, so you can compare your measurement with the exact same settings and really see the timing you get at the listening spot. Don't be too disappointed by the outcome, it might throw your whole theory down the drain...

the wavelet is the best, but it should be also a measrung which can the klippel do. and many other magazine or test sites use step response too, such filter examples you do many do here, but ignore filter and just measure speaker and speaker with subwoofer. then you notice that the step response get not longer with more bass and subwoofer. same as in my example.
this influence happen because of the phase shift or prering the filter do i think. nobody have here explain wy the step response reach after 9 ms 0 but the frequency the filter have is less than 50 hz. and 50 hz is period time of 20 ms. so wy the step response reach after 9 ms 0 and not at 20 ms or when it have 25 hz after 40 ms ?.

and if a speaker have a long step response then it is either the speaker slow or the speaker have a worse crossover that is steep or at low freq.

 

[Wesayso]

and if a speaker have a long step response then it is either the speaker slow or the speaker have a worse crossover that is steep or at low freq.

You couldn't be more wrong with this bit.... I gather your speaker doesn't have a "long step response" lol.

The STEP, IR and FR and Wavelet all show the same data. Just in a different manner. Look at it as another way of presenting the same data.
If you have the IR, you can calculate the others. But that doesn't mean you can make up what it all means like you are doing in this thread.

Learn by downloading RePhase and create some different IR's yourself to see it's relation with STEP and FR/Phase. Learn by applying some minimum phase filters and see what effect they have on the IR and STEP shape and form. Look at the Frequency response and Phase too!

A speaker that shows a STEP like I posted is the quickest one can have (and still be right), as all signals arrive at the exact same point in time.
Where it does deviate from that ideal is at the very top and bottom of it's spectrum, due to the roll off on both ends... As can be seen in the frequency graph.

Here's the STEP response if the signal is ranging from zero HZ to infinity:

It basically stays up forever....

Now lets bring in a "virtual" tweeter and woofer and make a Linkwitz Riley 4th order filter:

Possibly this graph above would be called fast in your definition, but actually, the tweeter (which still goes to infinity in this example) is ahead of the woofer (that still goes all the way down to 0 Hz). All that's seen here is the introduction of a (LR4) crossover.

Your posted examples clearly do not show an example of a true 'fast' speaker. It's just showing us that it has a crossover higher than a first order.

Now can we create a fast speaker with the help of some DSP? Sure, if you mind all bits and pieces that count, we can get quite close.

Here's a Wavelet measurement of my own (DSP corrected) speakers (Line arrays with a pair of Subwoofers) at the listening spot:

I used slightly different settings than the previous Waterfall here, but just look at the timing on the bottom axis.
It couldn't have resulted in this clean of a waterfall without room treatment though. Plus the type of speaker helps too.
Now that is the result of a fast speaker. I'll hope you show yours next .

The IR:

The shape of the STEP is entirely due to the specific room curve I have applied. Another factor that totally messes up how the STEP file looks to us. Even though this will probably be called "long" by you, it really is fast! As seen in the accompanying wavelet.

In this real world example, the high frequency arrives at the same moment as mid frequencies, all the way down to 300 Hz, where a bit of delay
is seen in the wavelet.

Now show your wavelet, is it really as fast as you think it is?

 

[bennybbbx]

It basically stays up forever....

Now lets bring in a "virtual" tweeter and woofer and make a Linkwitz Riley 4th order filter:

Possibly this graph above would be called fast in your definition, but actually, the tweeter (which still goes to infinity in this example) is ahead of the woofer (that still goes all the way down to 0 Hz). All that's seen here is the introduction of a (LR4) crossover.

Your posted examples clearly do not show an example of a true 'fast' speaker. It's just showing us that it has a crossover higher than a first order.

the step response is from the JBL 104 . the spec says it have a 6 db crossover.
see also my intro posts with the compare of kali and JBL 104 or other posts from me.

.
Now show your wavelet, is it really as fast as you think it is?

and which speaker you have ?. i have post my REW measure file . https://www.audiosciencereview.com/...r-itd-see-measures.28585/page-12#post-1029973 you can try yourself. but i do a screenshot and have open my settings windows. so you can see which settings i use. My phase is not so good as your measure but it look faster on many frequencies.

step response is a easy way to see and other speaker tests pages too does not show a wavelet.

 

[kyle_neuron]

your step response view does not show good the range from 0-1 ms. only this range is important for mid speed. i have upload rew another measure file, also do with a diffrent position. step response stay same. try yourself. linear phase filters do prering, this can also influence the step response. just try the test with a subwoofer and without subwoofer and do not change EQ and show the step response range from 0- 4 ms

I haven’t used any linear phase filters, in any of the plots. The results from EASERA are linear phase because the magnitude has been directly manipulated mathematically.

I have absolutely no clue what you’re on about with ‘mid speed’ in the first millisecond. Do you even realise that one period of 1 kHz is 1 millisecond long? What frequency range do you consider midrange to be?

You can keep moving the goalposts all you like to try and keep hold of your little theory, but you have yet to provide anything of worth to back it up.

By the way, a theory needs mathematical proof and a repeatable methodology. You haven’t come close to either.

Perhaps you might want to start over, and begin with understanding what your microphone type is.

 

[bennybbbx]

I have absolutely no clue what you’re on about with ‘mid speed’ in the first millisecond. Do you even realise that one period of 1 kHz is 1 millisecond long? What frequency range do you consider midrange to be?

with midrange i mean that the step response rise time and fall time should be as short as possible. this give good stereo width. then some people told that the decay time depend on Frequency range in low freq. this happen when use EQ. So i do as later post the example with the sub that the step response stay same if i use sub and much more lower frequency or not. I try also this example without highpass for the JBL. stay same

here can download the measured files.

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

For the third time. Both of your microphones are omnidirectional. This is stated by the manufacturers of both microphones themselves. Neither of them are ‘unidirectional’, as this would be a literal laser beam. Incorrect. Here is an example. I measured a single Faital Pro 3FE22 driver...

www.audiosciencereview.com

and here the screenshot that show that step response length stay same

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

??? The difference is there is a time delay and tiny amplitude difference between the signals picked up by the 2 mics. No difference in the frequency composition. with a time delay both stereo speakers can not play the same. when the girl sing at frequency 400 hz and it is at same time hear in...

www.audiosciencereview.com

the preview text that is seen is wrong. seem a forum problem. but when click on link show the correct post

 

[bennybbbx]

Perhaps you might want to start over, and begin with understanding what your microphone type is.

see this picture of the mems microphone i use. It have closed back and side. so it is not omnidirectional.

IK announces ARC System 2.5 with MEMS microphone

KVR Audio News: IK has announced the new ARC System 2.5 with MEMS microphone, which will be available for shipping in July 2017. ARC 2.5 is an acoustic correction system that combines a highly accurate MEMS measurem...

www.kvraudio.com

omnidirectional microphones get level from all sides with same level when i play a testone record with this microphone 6 khz then position to speaker to front spl ,meter show 62 db. on side and back 58 db

 

[Wesayso]

and which speaker you have ?

DIY line array speakers...

Even more info here: Thread on DIY audio

i have post my REW measure file . https://www.audiosciencereview.com/...r-itd-see-measures.28585/page-12#post-1029973 you can try yourself. but i do a screenshot and have open my settings windows. so you can see which settings i use.

You posted the wavelet, thanks... now learning to read it would be the next step.
You should see your tweeter is playing in front of your mid (about the 460 microseconds that the indicator shows).
That mid is clearly in front of any bass notes too.

As said, my own graph showed 300 Hz up to 20 KHz is firing instantly, no delay at all, all within the first fraction of a second.
If we look at 100 Hz -20000 Hz, it is all done within half a millisecond. No crossovers till we're down to 160 Hz in my speakers by the way.

My phase is not so good as your measure but it look faster on many frequencies.
step response is a easy way to see and other speaker tests pages too does not show a wavelet.

What part of your frequencies is faster exactly? How can it be faster, faster than what? If it's presenting a single line at 0, that's the reference.
That's where your tweeter is first in the race. So the mid's cannot be faster, as the delay can be seen in the wavelet.

From 2.5 KHz down to 200 Hz your mid response is lagging behind the tweeter by almost half a millisecond. So what was "so fast" about your speaker? That first narrow peak of the STEP is that tweeter, that was finished playing before the mid even starts. That's the true message your STEP is showing us. As is the Wavelet by the way. Just another view of the exact same information.

If you want to compare something, compare the frequency response, and phase if it's presented. Looking at an IR or STEP is almost
impossible to compare due to differences in crossovers used, crossover frequencies and room curves and possible room effects all
coming into play.
Even a wavelet, while being an easy graph to quickly see the large picture, will still hide some detail information, because as Kyle informed you: some frequencies only take a fraction of a millisecond to take place. Other, lower frequencies need more time from start to finish for one cycle.
All of that information is there, just not in the way you are 'reading' (into) it.

Here's your wavelet without sub:

And next up with the sub added:

(clearly the sub is late to the game, firing a full 40 ms after the main speaker)

And to remind us, what it "should" look like if everything was (at least close to) ideal:

So, after this information, as displayed here, shall we stop making uninformed claims about STEP responses being fast or slow?
What matters more (way more actually) is that both channels, left and right, have about equal timing and are close as far as
their frequency response is concerned. Then you get the best out of your Stereo.
Stop thinking you've got that STEP response all figured out, you're not even close! Re-read this entire thread and take to heart
all the good and wise information people wanted to show/give you.

I doubt my post will make you "see the light" but I just had to try. I've been secretly following/reading this thread for weeks and
each time I hope that one of us is able to "get trough" to you. This is my attempt. Lots of pictures to make it easier to follow.
And a real world result of my own speakers. Lots more in the thread over on DIY audio.

Hope it helps,

Ronald

 

[kyle_neuron]

see this picture of the mems microphone i use. It have closed back and side. so it is not omnidirectional.

IK announces ARC System 2.5 with MEMS microphone

KVR Audio News: IK has announced the new ARC System 2.5 with MEMS microphone, which will be available for shipping in July 2017. ARC 2.5 is an acoustic correction system that combines a highly accurate MEMS measurem...

www.kvraudio.com

omnidirectional microphones get level from all sides with same level when i play a testone record with this microphone 6 khz then position to speaker to front spl ,meter show 62 db. on side and back 58 db

Try reading the specification or documentation for something, rather than making yet another incorrect judgment from what you think you know.

I already linked this earlier but you obviously don't have any interest in reading anything people are saying or sharing.

Taken from the manufacturer's own website

ARC System 2 - Acoustic Correction System.

ARC System 2.5 is the first Acoustic Room Correction System plug-in with MEMS measurement microphone

www.ikmultimedia.com

Edit: my frustration is showing... sorry. I do try to at least engage and explain. Hopefully, you learn something...

The reason you get a different level response with a supposed omnidirectional pattern microphone at different orientations is caused by the fact the microphone has a physical body, which disturbs the pressure field.

There are 3 main types of measurement mic. Free field, diffuse field, and pressure field.

Most of the microphones you would buy for sound measurement as a consumer are the free field type. They're designed to work in - you guessed it - an open outdoor field, or anechoic environment like a chamber.

A free-field microphone capsule must be compensated to provide correct results in a pressure or diffuse field. A good microphone may have this done internally, either with electronics or physical design, or a compensation file is provided. An example of the latter is the isemCON EMX-7150, which comes with a compensation file for free field and diffuse field use.

http://www.micwaudio.com/uploaded/product/M-Series/Technical%20Note-%20Free%20and%20Pressure%20and%20Random%20Field%20Microphones.pdf

A diffuse field mic is a better choice for room acoustics with multiple noise sources - also known as a random incidence field. I use these when doing aerospace testing work, in reverb chambers.

Pressure field mics are what you'll find in your phone, or dummy heads. They're intended to be mounted in cavities that have the same complex pressure at any point.

Your MEMS mic is a free-field microphone. So is your Sonarworks XREF20, which I also have three of.

So now, let us consider your 'reference' headphone measurements. They are wrong, as you are using a free field microphone to measure a device which is designed to radiate into a pressure field - an ear cavity. Not only are you measuring the headphones without a proper representation of a head, but the mic is the wrong type.

Measurement Microphones Explained | Brüel & Kjær

In this article, we walk you through the physics of measurement microphones; from size, sensitivity and dynamic range to their frequency range and sound-field.

www.bksv.com

If your 'baseline reference' is wrong, and you don't know how your mic behaves, or the issues and limitations of your measurement methods, how can you claim that your data is accurate or correct?

More importantly, all measurements have a region of validity. If you’re generating a step response from data that’s not valid in the higher or lower frequencies, it will not be a correct or true representation of the speaker’s response.

This is why windowing is so important. It provides ‘anechoic’ data from a measurement of a diffuse field (a room) using a free field microphone. Likewise, knowing the frequency limits of data validity from ultra-near field and far field measurements.



I won’t even get into the fact that the well defined metric of Apparent Source Width has nothing to do with step response, and everything to do with radiation pattern and directivity, and the room’s reverberation tail.

 

[DonH56]

This has been going on in several threads since at least last July with numerous highly-qualified folk trying to explain the issue and what it all really means. The OP has made up his mind and is not about to change. Basic knowledge seems to be lacking and/or ignored. Waste of time posting, alas, except for posterity and the enlightenment of others.

 

[bennybbbx]

You should see your tweeter is playing in front of your mid (about the 460 microseconds that the indicator shows).

yes, this tweeter in front show the step response too. so step response is usefull to see if woofer and tweeter is time align good. later phase align did not work because in the crossover range where tweeter and woofer play together the diffrence of 460 µsec stay always in
but wy the sub should be 40 ms too late is realy strange. i guess room reflections or so. the kali stand behind monitor so no direct view to measure microphone. the kali have a lowpass at 48 hz. i delay the jbl with 7.5 ms. this are 2.5 meter or 8.4 feet the dsp show. the kali have 2-3 ms intern DSP delay . the kali as subwoofer is 1 meter behind the JBL. use step response come to the 7.5 ms delay thats need. i measure the kali delay with fullrange so the start can better measure. and the measured value of 7.5 ms are realistic. 40 ms delay mean that the woofer is 13 meter away or 45 feet. thats complete impossible. you can also look in step response of my REW dat, there can no peak see at 40 ms.

What part of your frequencies is faster exactly? How can it be faster, faster than what? If it's presenting a single line at 0, that's the reference.
That's where your tweeter is first in the race. So the mid's cannot be faster, as the delay can be seen in the wavelet.

look at 1 khz. on your picture there go the green range upto 2.5 ms. on my picture the green range go less than 1 ms. you see my settings. maybe you try display your speaker with my settings. results change alot depend on settings. to compare the wavelet need always same settings. and at 1 khz you need subtract the 460 µsec mid delay from my speaker. and then the green range from the mid is reach less than 0.5 ms.

 

[bennybbbx]

Try reading the specification or documentation for something, rather than making yet another incorrect judgment from what you think you know.

I already linked this earlier but you obviously don't have any interest in reading anything people are saying or sharing.

Taken from the manufacturer's own website

ARC System 2 - Acoustic Correction System.

ARC System 2.5 is the first Acoustic Room Correction System plug-in with MEMS measurement microphone

www.ikmultimedia.com

View attachment 175618View attachment 175619

they write this but i measure the level diffrence. should i do a video that show the level diffrence of 4 db in side and back ?. on low frequency of course every microphone is omnidirectional because there come much room reflections.

I have a sonarworks microphone. this is really omnidirectional but have more noise. my calibration file is ok. the mems microphone need lots of bass enhancement. this was the first time that i see that step response stay same even if i use microphone with diffrent FR.

On this post i have post calibration file of mems and sonarworks microphone

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

??? The difference is there is a time delay and tiny amplitude difference between the signals picked up by the 2 mics. No difference in the frequency composition. with a time delay both stereo speakers can not play the same. when the girl sing at frequency 400 hz and it is at same time hear in...

www.audiosciencereview.com

I guess this is freefeeld Freq correction. Or what do you think ?

you can see which diffrent frequency response the mems microphone have to the sonarworks XR 20 microphone. XR 20 have a SN and with this SN can download calibratiomn file

Pressure field mics are what you'll find in your phone, or dummy heads. They're intended to be mounted in cavities that have the same complex pressure at any point.

does phones not too use Mems chips because they are small ?

This is why windowing is so important. It provides ‘anechoic’ data from a measurement of a diffuse field (a room) using a free field microphone. Likewise, knowing the frequency limits of data validity from ultra-near field and far field measurements.

there is much between 0% or 100% possible which error rate do you think. note also i do compare measures. all my measure have same error, but this is enough to compare speakers or headphones when i have on speaker microphone 1.5 cm away.
and i can verify my FR with that whats measure on other page. and my results look and i think i have not more than 20% measure error to others.

I won’t even get into the fact that the well defined metric of Apparent Source Width has nothing to do with step response, and everything to do with radiation pattern and directivity, and the room’s reverberation tail.

 

[Wesayso]

Another big hint that you don't get what the pictures are trying to tell you.

The width of the area's isn't telling you anything, as indeed these are caused by differences in settings. Nothing less, nothing more.
If you are trying to determine how fast your speaker starts and stops, heck don't use a STEP response for that!
Filter your IR measurements if you want look at frequencies build up and stop again.

For timing the only thing you need to look at is that dotted line! So now it's about how fast the signal stops at different frequencies?
We need to introduce yet another tab of REW for that...

The Wavelet wouldn't be the best tool to show that, we need the waterfall.

Here's my waterfall plot in the first miliseconds of output:

And to show I'm not cheating, here's the plot from your REW measurement with the exact same settings:

I only adjusted the levels in-between the graphs to present about a 20 dB level drop. I't a bit harder to guess with your measurement.

Which speaker was faster, did you say? This is the stop side of the impulse, being generously filtered as a CSD plot for indoor measurements.
It would look worse if we turn that off. We want to see the speed right? Well, here it is in all it's glory.

Should I present an ideal picture to compare?

 

[bennybbbx]

Another big hint that you don't get what the pictures are trying to tell you.

The width of the area's isn't telling you anything, as indeed these are caused by differences in settings. Nothing less, nothing more.
If you are trying to determine how fast your speaker starts and stops, heck don't use a STEP response for that!
Filter your IR measurements if you want look at frequencies build up and stop again.

For timing the only thing you need to look at is that dotted line! So now it's about how fast the signal stops at different frequencies?
We need to introduce yet another tab of REW for that...

The Wavelet wouldn't be the best tool to show that, we need the waterfall.

Here's my waterfall plot in the first miliseconds of output:View attachment 175631

And to show I'm not cheating, here's the plot from your REW measurement with the exact same settings:
View attachment 175632

I only adjusted the levels in-between the graphs to present about a 20 dB level drop. I't a bit harder to guess with your measurement.

Which speaker was faster, did you say? This is the stop side of the impulse, being generously filtered as a CSD plot for indoor measurements.
It would look worse if we turn that off. We want to see the speed right? Well, here it is in all it's glory.

Should I present an ideal picture to compare?

your waterfall look much diffrent to the waterfall amirm do. i try your settings soon

The step response is create with laplace transform and integral(an area) of impulse response. you can read about laplace transform. it is used to see speed and lag in electronic or mechanical systems. speakers are mechanical and filters electronic.

 

[Wesayso]

This is a filtered waterfall to show the early signal. Otherwise all we'd see is the room response if we measure indoors without something nifty as the Klippel Amir is using. Stereophile often shows similar graphs of that very early signal. Driver plots often use similar settings.

The step response does show you all of that, but you'd need to know what you are looking at to interpret it. As long as we have all these different views available in a suite like REW, we can look at a more detailed picture of what we want to see.

You've pretty much proven the point that STEP responses are hard to read.
It is way too easy to read something into it that isn't right. If you want to know more about that STEP, just learn to use all other tools available within a suite like REW to see that part in more detail. That pretty much sums up the way I learned about this stuff. By looking at the information in every way possible. The Wavelet, the Waterfall, The frequency response with phase and things like group delay.

Comparing it to an ideal IR can be useful. To see how an ideal signal would behave and what it looks like under different settings and/or conditions.

Here's the ideal waterfall:

You see? My speaker isn't quite ideal either, but I did try hard to make it close....

 

[kyle_neuron]

Here's my waterfall plot in the first miliseconds of output:View attachment 175631

View attachment 175632

Oh, some fun!

Here's the 'main' speaker (Funktion One Res 4, by the way) from the measurements I posted, same settings:

And the wavelet:

With 1/24 frequency smoothing, because I'm no 'smooth criminal'.

Perhaps it's worth reminding people to check the y-axis scale too... your plots show about a 20 dB vertical range, so I tried to match that.

 

[Wesayso]

You have some nice toys to play with, @kyle_neuron

 

[bennybbbx]

The step response does show you all of that, but you'd need to know what you are looking at to interpret it. As long as we have all these different views available in a suite like REW, we can look at a more detailed picture of what we want to see.

maybe you can show step response with the range 0-5 ms too how it look ?.

If you want to know more about that STEP, just learn to use all other tools available within a suite like REW to see that part in more detail.

the newest tests from amir show now detail step response and there can see alot. . I have written that earlier. the reason wy step response is important to show is because it is 1. shown on many pages. to see details in REW feature from speaker tests there need the impulse response export as wav so can load in REW. but there is no impulse response and i can of course not buy 100 speakers and test them all in REW. so i need measures that help me to do a first choose of a speaker with fast mid. and for this i think the step response is the easiest way and available more often.

I also dont know what settings the klippel have for the waterfall. but currently there can only see the delay of DSP and resonances not if the mid driver is fast.

You see? My speaker isn't quite ideal either, but I did try hard to make it close....

as i read on your other links you use Vifa 9 BN 119/8 (europe name) wide band speakers. and this is my opinion. when can hear ITD the mostly used bass/mid woofers that can only reach 5-6 khz -3 db are not good enough for ITD hearing. so need on speaker test measures that show how much frequency the mid(/woofer reach. in tests can of course not open the speaker and measure the FR of the Bass/mid. so need a way to figure out of the existing speaker tests how fast the Bass/mid is and the time align of the systems. 3 way systems i not like because crossover at 250 hz or 400 hz sound not good to me. i do tests with kali and the jbl on top to simulate a 3 way system. for the dynaudio lyd 5 stand that crossover is 5300 hz. but look on amirm measurements the specs seem not true. crossover seem at 3.5 khz around. but how fast(which freq it reach -3 db) the bass/mid is not known.