Neumann KH 80 DSP Speaker Measurements: Take Two

NTK

Senior Member
Forum Donor
One question:

Why do you have to define where the acoustic axis beforehand, since it should be possible to calculate the frequency response in silico at any axis in the far field?
If you take a look at this post, the NFS scans in 2 concentric cylinders. That location of the speaker's reference axis doesn't matter much to the measurement points.
https://www.audiosciencereview.com/...-measuring-speakers.10725/page-17#post-301506

The reference axis should only be important for reporting purposes (i.e. the locations of the reporting points).

Juhazi

Major Contributor
Edge is just fine to demostrate this! Minimum mic distance is 1000mm, so I had to use a larger speaker than KH 80. In Edge we can't apply crossover so both drivers have ideal flat response to start with. Notice how interference changes! For other speakers just imagine different separation and distance, but basic mathematics is the same!

Thomas_A

Forum Donor
If you take a look at this post, the NFS scans in 2 concentric cylinders. That location of the speaker's reference axis doesn't matter much to the measurement points.
https://www.audiosciencereview.com/...-measuring-speakers.10725/page-17#post-301506

The reference axis should only be important for reporting purposes.
So then I don't understand why you should get a sloping response > 7 kHz in the far field just by redefining something beforehand when you should be able to do everything in silico. The original measurement of this speaker should, if everything is correctly calculated, look as the second measurement > 7 kHz, if you alter the axis a few centimetres down. But I guess this would not be the result, so then I am confused.

Pio2001

Active Member
Hi,
The most interesting result is this one :

It represents the correct balloon plot, but about 5 cm away from the speaker, not 2 meters.
And no it is not upside down. Since the reference point is taken below the tweeter. The measured sound in this point comes from above, thus, fires downwards.

The drop in high frequencies is consistent with the angle of the balloon plot : about -5 dB at 10 kz, which is what is expected about 35 or 40° from the axis, according to the other graph :

But... it is only consistent above 7000 Hz. As if the internal routines were using two different methods to calculate the frequency response above and below 7000 Hz.
Another possibility is a problem with microphone calibration. The frequency and amplitude anomaly is consistent with an uncalibrated microphone.

About the dip in low frequencies, it has been said above that the Klippel system should be more accurate than the others. However, let's not forget that low frequencies have large wavelength. Scanning two close surfaces, it should still be very difficult to separate direct sound from the first reflections.
In fact, below 100 Hz, we use to say that there is no such thing as a direct sound and a reflection. The whole wave is generated by the speaker and the walls at the same time.
True free field measurement will be the judge, if one day we can do it. Ground plane, in principle, should be a good reference too, but we are talking about 1 - 2 dB variations here. Is ground plane measurement accurate enough ?

NTK

Senior Member
Forum Donor
If the reference axis should be chosen before you start this Klippel spinorama, the system sucks....
I don't know what Klippel's software allows, but the NFS process doesn't need to know the location of the reference axis. In Klippel's patent application, they actually described a process of estimating the acoustical center of the speaker during the measurements (see item [0034]), so the location of the reference axis (if you based it on the acoustical center of the speaker) can be done automatically.

Thomas_A

Forum Donor
Hi,
The most interesting result is this one :

It represents the correct balloon plot, but about 5 cm away from the speaker, not 2 meters.
And no it is not upside down. Since the reference point is taken below the tweeter. The measured sound in this point comes from above, thus, fires downwards.

The drop in high frequencies is consistent with the angle of the balloon plot : about -5 dB at 10 kz, which is what is expected about 35 or 40° from the axis, according to the other graph :

But... it is only consistent above 7000 Hz. As if the internal routines were using two different methods to calculate the frequency response above and below 7000 Hz.
Another possibility is a problem with microphone calibration. The frequency and amplitude anomaly is consistent with an uncalibrated microphone.

About the dip in low frequencies, it has been said above that the Klippel system should be more accurate than the others. However, let's not forget that low frequencies have large wavelength. Scanning two close surfaces, it should still be very difficult to separate direct sound from the first reflections.
In fact, below 100 Hz, we use to say that there is no such thing as a direct sound and a reflection. The whole wave is generated by the speaker and the walls at the same time.
True free field measurement will be the judge, if one day we can do it. Ground plane, in principle, should be a good reference too, but we are talking about 1 - 2 dB variations here. Is ground plane measurement accurate enough ?
Since the ballon plots differs I can't see how that could be correctly calculated in the far field. You will get a sloping response in far field with the calculated data, but the actual response of the speaker is not like that.

Pio2001

Active Member
I forgot : I did some experiments moving my microphone in front of the tweeter of my KH-120 while playing a pink noise.
The ripples around 8000 Hz are there and depends on the microphone position. New ripples appear if I get my finger close to the microphone. They are just reflections on the microphone's body and stand.

Thomas_A

Forum Donor
I don't know what Klippel's software allows, but the NFS process doesn't need to know the location of the reference axis. In Klippel's patent application, they actually described a process of estimating the acoustical center of the speaker during the measurements (see item [0034]), so the location of the reference axis (if you based it on the acoustical center of the speaker) can be done automatically.
If so, why would Amir need to remeasure the speaker with respect to the discussion about acoustic axis? Just change the acoustic axis in silico and compare should be enough.

Last edited:

Pio2001

Active Member
Since the ballon plots differs I can't see how that could be correctly calculated in the far field. You will get a sloping response in far field with the calculated data, but the actual response of the speaker is not like that.
If I understand correctly what you are saying, that's also what I'm saying : these balloon plots are showing the extremely near field response of the speaker.
Something's wrong with the reference distance. Maybe it believes that we are asking for two inches instead of two meters.

DS23MAN

Member
Earlier on I mentioned the problem with using a free field calibrated measurement mic within a near field setup.

And this is the red herring. Because of the close range at exact tweeter axis on axis you will get a rise in the high frequencies ( source:Beyer Dynamic):

Attachments

• 128.7 KB Views: 87

Juhazi

Major Contributor
"We" wanted a new measurement for other reasons too. The main poit is that first one was so different from data publishd by the manufacturer and others. On reason for that might have been different 0-axis. And as the manufacturer has defined that to be in-between tweeeter and woofer, that should be used ( and the axis used must always be told as well as mic distance!)

Measurements don't mean a thing, if we don't know how they were done! By the way I am 30cm (12") long, but you guess where!

OP

amirm

Staff Member
CFO (Chief Fun Officer)
Did you notice the "Science" in the forum's name?
Science doesn't mean science project. What he stated is right on the money. The purpose of measurements for us is not to have graphs to stare at. They are there to predict what we may hear as far as speaker tonality. There, you need to remember the high level picture as I have explained over and over again. That is what the science says.

Juhazi

Major Contributor
Earlier on I mentioned the problem with using a free field calibrated measurement mic within a near field setup.

And this is the red herring. Because of the close range at exact tweeter axis on axis you will get a rise in the high frequencies ( source:Beyer Dynamic):
That is just why every good measurement mic comes with calibration file (usually different for axial and balloon/90deg). Amir said that he didn't bother to use it...

OP

amirm

Staff Member
CFO (Chief Fun Officer)
"We" wanted a new measurement for other reasons too. The main poit is that first one was so different from data publishd by the manufacturer and others. On reason for that might have been different 0-axis. And as the manufacturer has defined that to be in-between tweeeter and woofer, that should be used ( and the axis used must always be told as well as mic distance!)
No. You had no evidence that the manufacturer spec was correct in this regard. Something put in the manual is not proper data without back up. It is there to tell you to use your visual calibration at that point. Not for finding differences in measurements this way.

My sense is that they measured their speaker one way and put in a correction curve to get "flat" response relative to that measurement technique. That doesn't prove that the actual output of the speaker is that. It just says they have inverted the response to counter the variations they saw in that measurement.

Here is Klippel on issues with anechoic measurements:

You still seem to think that their data is the bible and anything found different must be an error on our part, and not theirs. Company's own measurement data is useless as it has just a graph and nothing else but the words "interpolated" written on it. The third-part measurements are very well subject to errors per their own documentation and issues stated above.

I am leaving room for NFS system having errors but you are a long way away from justifying the aggressive stance you are taking.

Your attack vector needs to be to get Neumann to prove their measurements are correct and are not affected by their measurement techniques.

Thomas_A

Forum Donor
If I understand correctly what you are saying, that's also what I'm saying : these balloon plots are showing the extremely near field response of the speaker.
Something's wrong with the reference distance. Maybe it believes that we are asking for two inches instead of two meters.
Yes, something like that. If you measure below the tweeter axis I would expect the balloon to be above the axis but not tilting down.

NTK

Senior Member
Forum Donor
If so, why you would Amir need to remeasure the speaker with respect to the discussion about acoustic axis? Just change the acoustic axis in silico and compare should be enough.
Again, I don't know what Klippel's software allows. Mathematically, I don't think Amir needs to redo his test for this purpose. But I think the original test was also done at a too high SPL.

Juhazi

Major Contributor
Science doesn't mean science project. What he stated is right on the money. The purpose of measurements for us is not to have graphs to stare at. They are there to predict what we may hear as far as speaker tonality. There, you need to remember the high level picture as I have explained over and over again. That is what the science says.
Then why the heck do you measure loudspeakers in the first place, if this difference is insignificant? Why are you worried, if a DAC has only 93dB SINAD? https://www.audiosciencereview.com/forum/index.php?threads/aimpire-ad10-fake-topping-d10-dac.11244/
---
(from post #1)

p.s. Do you have the explanation now? Has Wolfgang commented?

Last edited:
OP

amirm

Staff Member
CFO (Chief Fun Officer)
That is just why every good measurement mic comes with calibration file (usually different for axial and balloon/90deg). Amir said that he didn't bother to use it...
I don't bother because whether we are measuring speakers or rooms, that calibration is immaterial. The overall tonality of a speaker needs to be determined by taste. Whether it has 1 less dB more or less treble overall, is not material and something that is dealt with using a target curve in an EQ system.

What we care about are deviations from flat line, not overall slope.

If you take Klippel training, you see that all they have you put in is the sensitivity of the mic, not its frequency response variation.

Just a bit of absorption in your room will change the high frequency response way more than any microphone can measure.

Personal note: keep this style of posting going and I will learn my lesson to never bother dealing with you all's objections. Is it too hard to ask to you to be polite and constructive?