Dr. Klaus Heinz of HEDD Audio (ex ADAM Audio) - measuring speakers, in particular speaker dynamics

[ctrl]

the MTM have deeper bass as the LP6. and in this test there is no bass play. you hear only guitar and drums. but if you think its a bass problem i have do step response from 300 hz upto 3 khz. the results are the same. lp6 is lots slower. see setting for measure. it play only sine wave from 300 hz to 3 khz.

The distance between the tweeter and woofer peaks in the step response is simply a function of the crossover frequency and the filter order.

The higher the crossover frequency, the closer the tweeter and woofer peaks (in the step response) are together.

The higher the filter order at the crossover frequency, the greater the phase shift and thus the spacing of the peaks (in the step response).

The LP6 has a crossover frequency of 1.5kHz, the MTM one of 3.1kHz - this alone explains the differences.

In addition, the MTM offers a "linear-phase crossover at 3.1 kHz with variable edge steepness", which eliminates the phase shift at the crossover frequency and "saves" a few fractions of ms.

If the LP6 also had a crossover frequency of 3.1kHz, the difference in the tweeter and woofer peaks would be only minimally different from the MTM.

What you think you perceive as a difference in sound is most likely due to other causes, such as FR and directivity.

 

[bennybbbx]

The distance between the tweeter and woofer peaks in the step response is simply a function of the crossover frequency and the filter order.

The higher the crossover frequency, the closer the tweeter and woofer peaks (in the step response) are together.

The higher the filter order at the crossover frequency, the greater the phase shift and thus the spacing of the peaks (in the step response).

The LP6 has a crossover frequency of 1.5kHz, the MTM one of 3.1kHz - this alone explains the differences.

In addition, the MTM offers a "linear-phase crossover at 3.1 kHz with variable edge steepness", which eliminates the phase shift at the crossover frequency and "saves" a few fractions of ms.

If the LP6 also had a crossover frequency of 3.1kHz, the difference in the tweeter and woofer peaks would be only minimally different from the MTM.



What you think you perceive as a difference in sound is most likely due to other causes, such as FR and directivity.

so it is maybe bad that the crossover frequency of Kali is 1.5 khz ?. most speakers use crossover over 2.3 khz. the LP6 is wider in directivity but sound much smaller
https://audioxpress.com/article/testing-loudspeakers-which-measurements-matter-part-2
here is explain about step impulse too. the impulse should be fast and short.there is example how ideal speaker look. see headphone impule. A 8 eur headphone with small membrane but good bass reach this when measure from 40 hz.

AKG k1000 have larger membrane, this look more as speaker result but still much better as speakers

 

[ctrl]

so it is maybe bad that the crossover frequency of Kali is 1.5 khz ?. most speakers use crossover over 2.3 khz.

As long as the crossover frequency and filter order keep the resulting group delay well below the audibility limit, there is nothing at all "bad" or particularly "good" about a low or high crossover frequency (if you exclude other aspects).
Unless you believe that you can perceive group delay in loudspeakers even lower than the limits of perceptibility determined with artificial signals via headphones - in this case, there are no more scientific arguments.

Source: wikipedia

If an ideal 2-way loudspeaker has a crossover frequency of 500Hz, then the step response "looks quite bad" - but this is simply due to the low crossover frequency.
The resulting group delay is still far below the perception threshold.

A 8 eur headphone with small membrane but good bass reach this when measure from 40 hz.

An ideal full-range loudspeaker/headphone that transmits sound ideally from 20-20kHz has the following step response:

Therefore, something is wrong with your first example. The step response of the AKG looks reasonable.

 

[bennybbbx]

I dont know what should be wrong in the panasonic step response measure. i do it tomorrow again. i do it several times before. same result. I have also more headphones i can measure. if you think the K1000 look reasonable. the curve go down after 0,25 ms to 20% level. the MT reach after 1 ms 20% level. the LP6 go after 1.6 ms to 20% level. The KH 120 after 1.5 ms . the spectrasonic S8 after 0.9 ms .there is lots diffrence. and in theory it is clear that a speaker that need 1 ms or more can not produce exact transient over 1 khz. as my stereo width example show seem some or more does not hear it. but in forum are only few and all have not test, so can not say how much can hear the extrem smaller stereo width from kali in compare to MTM or Canton

here is group delay of MTM and Kali. MTM have a large group delay peak at 300 hz. whats this ?. but all in all kali group delay look nicer ?.

 

[ctrl]

here is group delay of MTM and Kali. MTM have a large group delay peak at 300 hz. whats this ?. but all in all kali group delay look nicer ?.

During the measurement, the loudspeaker should be placed as freely as possible in the room, with maximum distance to boundary surfaces (walls, floor, ceiling, mixing console, table...). The evaluation of the measurement should always be done via a gate to eliminate reflections. Since it is not important to measure the FR as accurately as possible, you can measure at a distance of 0.3m.

Read for REW how to perform a gated measurement, then there will be no more outliers in the group delay due to reflections.

 

[j_j]

The rise time that elapses until the tweeter peak, for example, is a measure of the frequency response limit of the tweeter. Your example shows a step response with an infinite frequency limit.

That's only true if there is no excess phase in the system.

 

[bennybbbx]

During the measurement, the loudspeaker should be placed as freely as possible in the room, with maximum distance to boundary surfaces (walls, floor, ceiling, mixing console, table...). The evaluation of the measurement should always be done via a gate to eliminate reflections. Since it is not important to measure the FR as accurately as possible, you can measure at a distance of 0.3m.

Read for REW how to perform a gated measurement, then there will be no more outliers in the group delay due to reflections.

yes this is clear. But if the room really have a influence there must be simular peek and valeys. see frequency response measure here.

at 1 min 26. You can see in bass range look very simular the both speakers. so it is clear much influence have the room. but on group delay see much diffrence. The Canton with the Kali as subbass look diffrent too. only on frequency below 100 hz they look simular

EDIT: I do 3 measurements with the panasonic headphone with diffrent volume and microphone position. the imnportant piece is always same.

and here another headphone trust G322c but have larger membrane as panasonic and smaller membrane as AKG. also measure with 3 diffrent volumes and microphone positions. but can see clear the small panasonic have faster impulse response

. frequency response can see HRTF curve all headphones have i can not reduce.

 

[bennybbbx]

In REW the overlays did not show step response. only impulse. here is measure of the trust headphone as step response

 

[ctrl]

That's only true if there is no excess phase in the system.

Yep, thanks for the reminder.

When considering an ideal tweeter, the excess phase does not play a role (since a minimum-phase system is assumed there), but of course it does in real measurements.
There, the excess phase influences the course of the step response and makes its interpretation even more difficult.

 

[j_j]

Yep, thanks for the reminder.

When considering an ideal tweeter, the excess phase does not play a role (since a minimum-phase system is assumed there), but of course it does in real measurements.
There, the excess phase influences the course of the step response and makes its interpretation even more difficult.

And when you use impulse response, it's easier to identify time delay. And, just speaking from experience, most tweeters, even really good ones, are not exactly minimum phase. You can see that delay across most any dome, more or less, especially if you measure 3 points (0, +30,0 and -,+30) in the radiation pattern. A good ribbon looks more like a moving membrane, so you see loss in the vertical, instead.

 

[tuga]

The problem is that "microdynamics" and "macrodynamics" are audio reviewer terms, not things with an actual definition.

It is not difficult to measure dynamic range, linearity, power compression, and similar characteristics, but that's not a good story to tell the folks who influence your buyer base.

Shouldn't we also be looking at measuring low-level accuracy or linearity?

 

[SIY]

Shouldn't we also be looking at measuring low-level accuracy or linearity?

OMG YOU FOUND THE FLAW!!!!

Umm, you do understand that we routinely measure to -150dB? Do you have any idea of how "low level" that is?

 

[tuga]

OMG YOU FOUND THE FLAW!!!!

Umm, you do understand that we routinely measure to -150dB? Do you have any idea of how "low level" that is?

Speaker distortion at -150dB? Wow, you're good.

I though that this topic was about speakers. The title is "measuring speakers, in particular speaker dynamics". And it's in the "Room Acoustics and General Speaker Discussions".

 

[Killingbeans]

 

[andreasmaaan]

Shouldn't we also be looking at measuring low-level accuracy or linearity?

I do think understanding a loudspeaker's nonlinear distortion profile at low SPLs is important.

The best transducers produce low distortion at low SPLs, with distortion of course rising as SPL rises. Many transducers, however, produce relatively high levels of distortion (especially odd-order distortion) at low SPLs, too. If nonlinear distortion measurements are taken at only one (relatively high) SPL, a transducer or loudspeaker's low-level distortion may be missed.

Since our ability to hear distortion actually increases as SPL decreases, it is this lower-level distortion that may actually be more important in terms of audibility.

 

[SIY]

Speaker distortion at -150dB? Wow, you're good.

I though that this topic was about speakers. The title is "measuring speakers, in particular speaker dynamics". And it's in the "Room Acoustics and General Speaker Discussions".

If a speaker has distortion that low, we can see it. No speaker does, and no chamber is that quiet, but our distortion measurement can and does. You’re digging in a dry hole.

 

[tuga]

If a speaker has distortion that low, we can see it. No speaker does, and no chamber is that quiet, but our distortion measurement can and does. You’re digging in a dry hole.

Why would anyone measure speaker distortion at -150dB to full scale and how loud would that full scale be anyway, 170dB?

 

[SIY]

Why would anyone measure speaker distortion at -150dB to full scale and how loud would that full scale be anyway, 170dB?

You asserted the lack of low level measurements. Make up your mind.

 

[andreasmaaan]

It seems to me there is a cross-purposes argument going on here.

When you said “low-level” @tuga, I took you to mean testing with low-level signals (as opposed to testing for low-level distortions relative to high-level test signals).

Was that interpretation correct?

 

[tuga]

It seems to me there is a cross-purposes argument going on here.

When you said “low-level” @tuga, I took you to mean testing with low-level signals (as opposed to testing for low-level distortions relative to high-level test signals).

Was that interpretation correct?

Yes, that is what I meant.