Neumann KH 80 DSP Speaker Measurements: Take Two

[ctrl]

Hmm, looking at this video

Thanks to gr-e for the link, it really helps to understand the measurements made.

At 06:28 are this the settings we have discussed before? The "tweeter point" and the "reference point". The "reference point" would be the reference point given by the manufacturer - or is this something else again?

At 06:28 in the video you can see the reason for the "strange" cumulative spectral decay diagrams posted by Amir.
As default setting for the impulse response analysis the beginning of the impulse is selected. To get a usable CSD the evaluation should start a few samples before the impulse (the exact value is determined by trial and error).
In detail I have described this here.

At 13:09 we see that the default radius for the evaluation is probably set to 10m, think Amir will have set it to 2m.
But I don't know if these settings are also used for the evaluation according to CEA 2034.

 

[gr-e]

At 06:28 are this the settings we have discussed before? The "tweeter point" and the "reference point". The "reference point" would be the reference point given by the manufacturer

Yeah. The tweeter point is the origin of the "internal coordinate system" and the reference point is the origin of "standard coordinate system" (that is used for data presentation), as mentioned on this slide I posted earlier:
>Full doc

 

[edechamps]

At 13:09 we see that the default radius for the evaluation is probably set to 10m, think Amir will have set it to 2m.

I'm a bit confused as to what the radius means, actually. Is it only used to scale the SPL or does it change the computed sound field? If it's the former, then it's a bit confusing if one sets it to, say, 1 meter, and the result doesn't match what the sound field actually looks like at 1 meter. If it's the latter, then it's also confusing because it would end up computing the near field in a control panel tab that is literally called "far field".

 

[gr-e]

Yeah. The tweeter point is the origin of the "internal coordinate system" and the reference point is the origin of "standard coordinate system" (that is used for data presentation), as mentioned on this slide I posted earlier:
View attachment 49579 >Full doc

I'm not sure where exacly the reference point is set, maybe here in the "setup points" section:

 

[amirm]

FYI, there are three automatic settings for reference point. The one it uses, the max and mean SPL. I tried all three and they all generate the identical response for this speaker. There are two other ones for manual entry but they require figuring out what to put in there. I have not had time to do that.

 

[ctrl]

You guys want to get Neumann to come here and defend their measurements and explain how their speaker is free of baffle diffraction that causes such dips on-axis, I am game. Otherwise we are back down the unhealthy path of obsessing over detail that is not material. And exists in every type of measurements.

Amir, take a closer look at these two anchoic chamber readings.
First, the scale of the y-axis in the diagram is already "finer" with 40dB than the usual 50dB.

In addition, the blue curve definitely does not use smoothing, but the black curve does have a slight smoothing. This should slightly amplify the outlier at 5kHz.

Then look at the really important frequency range 100Hz to 10kHz of the two anchoic chamber measurements. The deviation of the measurements is about 1dB (with a few exceptions). With some smoothing (1/24oct or 1/12oct) the deviations should average out even further. Nobody here will expect better.
About the possible accuracy of DIY measurements (about +-1dB) compared to measurements in an anechoic chamber, I wrote here.

explain how their speaker is free of baffle diffraction that causes such dips on-axis

Nowadays it is not so difficult to reduce the baffle diffraction to a minimum, so that the off-axis frequency responses run parallel (with the on-axis response) over wide frequency ranges and do not show discontinuities due to baffle diffraction.

Here is an example of a 0.25m wide passive loudspeaker. The "biggest" effects of the baffle diffraction should show up in the range 2-3.5kHz (which is still slightly present) - symmetrical driver arrangement, measurements horz. 0-90deg:

In the frequency range 1kHz to 10kHz the measurements show good parallelism up to almost 60°.

 

[edechamps]

FYI, there are three automatic settings for reference point. The one it uses, the max and mean SPL. I tried all three and they all generate the identical response for this speaker. There are two other ones for manual entry but they require figuring out what to put in there. I have not had time to do that.

Yep. I think it would make sense to try with the manufacturer-specified reference axis, which in the case of the KH 80 is 52 mm below the twitter axis, if you can find the time. My expectation is that this would reduce the treble tilt, bringing the overall trend of the high frequency response more in line with third party measurements.

 

[gr-e]

Yep. I think it would make sense to try with the manufacturer-specified reference axis, which in the case of the KH 80 is 52 mm below the twitter axis, if you can find the time. My expectation is that this would reduce the treble tilt, bringing the overall trend of the high frequency response more in line with third party measurements.

I don't think there will be much difference, it's only a 1.5 degree change @2m.
That said, I agree that manufacturer's reference axis should be used for future measurements, and if it's not specified the report should mention which axis was used. (As long as Amir has time and energy for all this )

 

[KSTR]

At some realistic listening distance a small offset of the reference axis is effectively irrelevant, the vector direction is what is relevant (normal to the baffle, in most cases). But for the near-field measurement procedure it is not, as we have learned by now. That's why the tweeter reference point has to be spot on.

EDIT:
Looking at the KH80 manual, Neumann doesn't specify the offset, only the direction, @edechamps. I think the image was mistaken by us because it shows the vector origin between woofer and tweeter, for convenience of drawing, not for showing a specified position on the baffle where the vector references to.

The acoustical axis is a line perpendicular to the loudspeaker’s front panel along which the microphone was placed when tuning the loudspeaker’s crossover during design. Pointing the acoustical axis, in the horizontal and vertical planes, towards the listening position or center of the monitoring area will give the best measured and perceived sound quality.

 

[amirm]

The other question is how the reference point would be below the tweeter as frequencies go up above the crossover. All the sound would be produced by the tweeter so by definition that is the reference axis, not what is below it.

 

[MZKM]

The other question is how the reference point would be below the tweeter as frequencies go up above the crossover. All the sound would be produced by the tweeter so by definition that is the reference axis, not what is below it.

Well, on the tweeter axis it has some excess upper treble, going slightly below would tame that.

All this being said, with this much difficulty understanding how to ideally measure this speaker, just think about the people owning this and deciding how to set it up.

The reason you saw such a tilt-down while using the manufacturer’s stated reference axis is because while correct in the vertical plane, you are measuring so close that the relative degree off-axis is much larger that at the same vertical height but measuring at ~4ft out, which is in the recommended listening distance.

The vertical off-axis around the crossover region didn't really change compared to the tweeter vs reference axis. So, why on Earth did they make the mid-point the reference axis? The world may never know.

 

[gr-e]

The other question is how the reference point would be below the tweeter as frequencies go up above the crossover. All the sound would be produced by the tweeter so by definition that is the reference axis, not what is below it.

The reference axis is just as relevant at crossover frequency as it is at higher frequencies because it defines how woofer and tweeter interact.
Why did manufacturer choose the point between woofer and tweeter? I guess because the relation between driver phases remains the same along this axis and you will get the least variation in FR at different distances.

 

[KSTR]

Looking at the KH80 manual, Neumann doesn't specify the offset, only the direction, @edechamps.

Oops, I have to correct myself. In the version of the manual that I just downloaded now the actually *do* specify the reference point exactly in between woofer and tweeter. In old preliminary german datasheet I had on my harddrive they didn't

 

[napilopez]

At some realistic listening distance a small offset of the reference axis is effectively irrelevant, the vector direction is what is relevant (normal to the baffle, in most cases). But for the near-field measurement procedure it is not, as we have learned by now. That's why the tweeter reference point has to be spot on.

EDIT:
Looking at the KH80 manual, Neumann doesn't specify the offset, only the direction, @edechamps. I think the image was mistaken by us because it shows the vector origin between woofer and tweeter, for convenience of drawing, not for showing a specified position on the baffle where the vector references to.

Am I misunderstanding? The offset is in the manual. 12.9cm from the bottom of the speaker.

They also specify how they define "acoustical axis," or what we've been calling the "reference axis." It is both how the speaker is meant to be positioned to measure and sound best. Of course, small movements will affect your placement, but it's still better for those movements to be relative to the reference axis than some other point.

"The acoustical axis is a line normal to the loudspeaker’s front panel along which the microphone was placed when tuning the loudspeaker’s crossover during design. Pointing the acoustical axis, in the horizontal and vertical planes, directly at the listening position or centre of the listening area will give the best measured and perceived sound quality. "

EDIT: Nevermind @KSTR , just saw your most recent post =]

 

[KSTR]

The vertical off-axis around the crossover region didn't really change compared to the tweeter vs reference axis. So, why on Earth did they make the mid-point the reference axis? The world may never know.

My guess is that this position gives better results when actually measuring close. The all important driver summation at XO freq will not be severly compromised whereas the slightly reduced top end is much less a concern.
I for one also always used the midpoint between woofer or midrange and tweeter for specifications.

 

[ctrl]

So, why on Earth did they make the mid-point the reference axis? The world may never know.

At this point, or better on this axis, the sound travel time of tweeter and woofer is the same - if the depth offset on the baffle of tweeter and woofer is the same, which is probably achieved with the waveguide for the tweeter.
So the relative difference in sound travel time is always zero.

This means that during development there is no additional phase shift due to different propagation times of tweeter and woofer, which would lead to an inclination of the radiation lobe in the vertical axis at the crossover frequency.

 

[amirm]

All this being said, with this much difficulty understanding how to ideally measure this speaker, just think about the people owning this and deciding how to set it up.

With two ears and a person moving around, none of this precision is required or practical.

Here is the bottom line: I will continue to measure at tweeter axis and 0 degrees to the baffle. Measurements are provided for off-axis radiation and people can choose to use those and align speakers accordingly.

 

[MZKM]

With two ears and a person moving around, none of this precision is required or practical.

Here is the bottom line: I will continue to measure at tweeter axis and 0 degrees to the baffle. Measurements are provided for off-axis radiation and people can choose to use those and align speakers accordingly.

No real complaints here.

10° increments should be enough to someone to get the picture (compared to 5°).

 

[Haint]

Well, on the tweeter axis it has some excess upper treble, going slightly below would tame that.

All this being said, with this much difficulty understanding how to ideally measure this speaker, just think about the people owning this and deciding how to set it up.

The reason you saw such a tilt-down while using the manufacturer’s stated reference axis is because while correct in the vertical plane, you are measuring so close that the relative degree off-axis is much larger that at the same vertical height but measuring at ~4ft out, which is in the recommended listening distance.

The vertical off-axis around the crossover region didn't really change compared to the tweeter vs reference axis. So, why on Earth did they make the mid-point the reference axis? The world may never know.

Yea this seems to be the most logical explanation for my money.

 

[bobbooo]

There is still the dip >13kHz in sample #2, which is likely in the speaker itself and not due to the measurement.

If this is the case, and this dip is due to speaker unit variation, then the average of the preference scores for the two units measured should be used in your scoring charts, as this would give a more accurate prediction of the performance of any particular unit a customer might receive. So the score for the Neumann should be 5.29, which makes it pretty much joint best speaker so far with the Harbeth (5.31).