• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Understanding How the Klippel NFS Works

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
I also saw some comments from other people.
Only a handful of people own/understand its operation. So if you heard that, it is an uniformed opinion
  • NFS is the most inaccurate among common electroacoustic testers.
Just as I said above.
  • It has a protective cover for that microphone, the high frequency is easy to be wrong, take off the protective cover, if the setting is wrong, the results will be wacky.
Not at all. There is no cover for the microphone. There is a mount for the microphone that is about 2 inches or so. This can cause very slight error when its dimension matches that wavelength. I removed mine and put padding on boom fixture shortly after I got the system. This completely solved the small reflections.

Note that the above has nothing to do with NFS itself as the same issue would exist with standard measurements.

  • NFS is a cost down thing, the anechoic chamber is too expensive.
Nope. As explained by NTK, me and Klippel documentation, the system is far superior to anechoic chambers used for audio. The low frequency response is absolutely superior and its resolution is one degree which is far, far better even fine grained anechoic measurements. Even companies like Harman that have more than one anechoic chamber, also own Klippel NFS.

The advantage of anechoic room is speed. There is no computational phase. What you measure is what you get. And if you build a microphone array and use a turntable, you can get your 10 degree measurements much faster. It can also be extremely quiet (if built that way) allowing measurements of THD+N that would go very low.

  • I remember that the high frequency of NFS is filtered by the time window, which is different from the anechoic chamber in principle. Mainly low frequency and SNR of the algorithm.
Gated response gives you perfect anechoic above 1 kHz or so. The reflection is completely removed by the time slice. The problem with it is that you lose low frequency resolution which Klippel solves with field separation.
  • The microphone for NFS is relatively low-end GRAS.
This is complete nonsense. NFS doesn't even come with a microphone. You can buy/use any you want. Klippel does resell some other microphones from GRAS and Microtech Geffell. I own the latter which is an incredible microphone with very low distortion. But again, you can use any microphone and enter its calibration into the system.

My NFS measurements have been compared to anechoic measurements by both Genelec and Neumann and they are extremely well correlated. The slight difference between mine them has no answer: that is, whether they are wrong or I am. It is very small and only in bass where room impacts anyway.

Here is my measurements of Neumann KH80DSP:

index.php


And here is company's anechoic chamber measurements:
index.php


It doesn't get any better than this. Or better proof of your FUD in this regard.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
The influence of the edge diffraction of the speaker on the sound field distribution cannot be considered. Edge diffraction still has no small effect on far-field acoustic radiation.
Diffraction effects are absolutely included in the measurements of NFS. No idea what you are saying or claiming here.
 

Curvature

Major Contributor
Joined
May 20, 2022
Messages
1,074
Likes
1,353
KH80 Comparison.png

Three quality sources Neumann (full anechoic probably), ASR (NFS) and S&R (hemi anechoic plus nearfield for bass) showing KH80 within 1 or 2dB on axis.
 

192kbps

Addicted to Fun and Learning
Joined
Apr 3, 2019
Messages
534
Likes
614
Only a handful of people own/understand its operation. So if you heard that, it is an uniformed opinion

Just as I said above.

Not at all. There is no cover for the microphone. There is a mount for the microphone that is about 2 inches or so. This can cause very slight error when its dimension matches that wavelength. I removed mine and put padding on boom fixture shortly after I got the system. This completely solved the small reflections.

Note that the above has nothing to do with NFS itself as the same issue would exist with standard measurements.


Nope. As explained by NTK, me and Klippel documentation, the system is far superior to anechoic chambers used for audio. The low frequency response is absolutely superior and its resolution is one degree which is far, far better even fine grained anechoic measurements. Even companies like Harman that have more than one anechoic chamber, also own Klippel NFS.

The advantage of anechoic room is speed. There is no computational phase. What you measure is what you get. And if you build a microphone array and use a turntable, you can get your 10 degree measurements much faster. It can also be extremely quiet (if built that way) allowing measurements of THD+N that would go very low.


Gated response gives you perfect anechoic above 1 kHz or so. The reflection is completely removed by the time slice. The problem with it is that you lose low frequency resolution which Klippel solves with field separation.

This is complete nonsense. NFS doesn't even come with a microphone. You can buy/use any you want. Klippel does resell some other microphones from GRAS and Microtech Geffell. I own the latter which is an incredible microphone with very low distortion. But again, you can use any microphone and enter its calibration into the system.

My NFS measurements have been compared to anechoic measurements by both Genelec and Neumann and they are extremely well correlated. The slight difference between mine them has no answer: that is, whether they are wrong or I am. It is very small and only in bass where room impacts anyway.

Here is my measurements of Neumann KH80DSP:

index.php


And here is company's anechoic chamber measurements:
index.php


It doesn't get any better than this. Or better proof of your FUD in this regard.

Thank you as always, because a former Harman employee from our country made several speakers after quitting, and when I asked him if he could provide NFS measurements for the speakers, He asked me to provide evidence, both theoretically and experimentally, that NFS can be superior to an anechoic chamber + AP, and made those points, I had to resort to forums, Sorry, thank you again for your patient reply.


index.php


I also found this thread, may I ask if this is because of the microphone? or other unknown factors.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
index.php


I also found this thread, may I ask if this is because of the microphone? or other unknown factors.
It was a factor of temperature. And slightly due to microphone fixture reflection.
 

mwmkravchenko

Active Member
Joined
Nov 20, 2018
Messages
180
Likes
187
Location
Perth Ontario
Everything that used to make measurements is a tool. Tools have drawbacks faults and limitations. I have a large yard, live out in the countryside and own very tall ladders. So I have the ability over parts of the year to do true anechoic measurements and obstruction free ground plane measurements. But. In the winter, 5 months of the year here near Ottawa, I need another solution. If I can make comparisons between the two measurement methods I have a means of calibrating one versus the other type of measurement. Regarding Mics I have 13 mics. They all have a little difference between them, or limitations. The 1 inch mics are good to 177db but limited to about 8 kilohertz. 1/2 inch about 18khetz, 1/4 inch near 40Khertz. In common use they are closer in their response than my microphone positioning errors. Even the UMIK is spot on as long as you are not overloading it. And it is portable and easy to use. Plugs into my phone and Audio tool can be use don site as a generator and measurement setup. So we work with what we have. The tools are not the key to understanding what the measurements are. Nor are they the means to getting adequate measurements. I am guessing you know all of this. Your comments show a great amount of understanding of these ideas.
 

alex-z

Addicted to Fun and Learning
Joined
Feb 19, 2021
Messages
910
Likes
1,684
Location
Canada
It was a factor of temperature. And slightly due to microphone fixture reflection.

Is it possible to temperature correct the data via the Klippel software, or is that fixed at the time of starting the sweeps?

Just curious because I have been comparing some of my own data (Q100 and BS22) vs the Klippel NFS, and hadn't considered what would happen with a swing of 10-15 degrees.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Is it possible to temperature correct the data via the Klippel software, or is that fixed at the time of starting the sweeps?
Not for any specific driver/speaker. Note that this is a hypothesis. As I mentioned neither one of us knows which version is right as we don't have a gold standard to compare to. At at any rate, acoustic measurements are never meant to be scrutinized at this level.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,421
Likes
2,406
Location
Sweden
Perhaps answered already somewhere, but is there any way, from Klippel data, to recalculate the response using two speakers in the classic stereo triangle?
 

edechamps

Addicted to Fun and Learning
Forum Donor
Joined
Nov 21, 2018
Messages
910
Likes
3,620
Location
London, United Kingdom
Perhaps answered already somewhere, but is there any way, from Klippel data, to recalculate the response using two speakers in the classic stereo triangle?

Assuming you meant "two identical speakers", your question doesn't seem to make sense. NFS data is anechoic data. The summed anechoic response of two identical speakers is just the response of one speaker doubled (i.e. +6dB). Obviously that's not very useful.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,421
Likes
2,406
Location
Sweden
Assuming you meant "two identical speakers", your question doesn't seem to make sense. NFS data is anechoic data. The summed anechoic response of two identical speakers is just the response of one speaker doubled (i.e. +6dB). Obviously that's not very useful.
I was not very clear sorry. It would include HRTF as well.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,421
Likes
2,406
Location
Sweden
You can pull anechoic data from Klippel and take it to matlab or some other program to post process what happens with two speakers playing. By itself it can't do that.
Thanks for info. I guessed that another software was needed but I was not sure.
 

Koeitje

Major Contributor
Joined
Oct 10, 2019
Messages
2,292
Likes
3,880
View attachment 214688
Three quality sources Neumann (full anechoic probably), ASR (NFS) and S&R (hemi anechoic plus nearfield for bass) showing KH80 within 1 or 2dB on axis.
Most likely just the difference between the actual loudspeakers, not the measurement. No two loudspeakers are exactly the same, and Neumann is one of the best in the market when it comes to this. You will see much worse difference for many other speakers.
 

Curvature

Major Contributor
Joined
May 20, 2022
Messages
1,074
Likes
1,353
Most likely just the difference between the actual loudspeakers, not the measurement. No two loudspeakers are exactly the same, and Neumann is one of the best in the market when it comes to this. You will see much worse difference for many other speakers.
I don't think we can make that conclusion. The differences are too small and we don't have enough data.
 

YSC

Major Contributor
Joined
Dec 31, 2019
Messages
3,194
Likes
2,570
recently reading the D&D 8C distortion measurement from Erin, just wonders how the NFS did the distortion, was it facing the acoustics axis and did a sweep, or was it doing the 360 scan and compose the distortion via math? if it's the letter I am curious about will it create extra distortion figures due to the cardiode design
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
recently reading the D&D 8C distortion measurement from Erin, just wonders how the NFS did the distortion, was it facing the acoustics axis and did a sweep, or was it doing the 360 scan and compose the distortion via math? if it's the letter I am curious about will it create extra distortion figures due to the cardiode design
Klippel NFS itself does not include specific distortion measurement capabilities. Both Erin and Amir do distortion measurement in on-axis nearfield.

The increased harmonic distortion of the D&D below 200Hz is most likely caused by the cabinet design with the side slots and the low crossover frequency of the midwoofer at 100Hz to the rear subwoofers - in case you are wondering about this.
This is why the HD drops a bit below 100Hz when the subwoofers play alone.
 

mwmkravchenko

Active Member
Joined
Nov 20, 2018
Messages
180
Likes
187
Location
Perth Ontario
Klippel NFS itself does not include specific distortion measurement capabilities. Both Erin and Amir do distortion measurement in on-axis nearfield.

The increased harmonic distortion of the D&D below 200Hz is most likely caused by the cabinet design with the side slots and the low crossover frequency of the midwoofer at 100Hz to the rear subwoofers - in case you are wondering about this.
This is why the HD drops a bit below 100Hz when the subwoofers play alone.
I would argue that it is coming from the forward facing driver. Many drivers are starting to have distortion in this area of reproducing music as this is an area in the upper base that has a lot of energy no matter the genre of music. Kick drum is near 120 hertz as an example. Chest cavity resonance is usually 117hz. Where to trade off the duties from driver to driver is not always as simple as picking a crossover point. And I think this is a good case to learn from. A great loudspeaker that could see some improvement.
 

YSC

Major Contributor
Joined
Dec 31, 2019
Messages
3,194
Likes
2,570
Klippel NFS itself does not include specific distortion measurement capabilities. Both Erin and Amir do distortion measurement in on-axis nearfield.

The increased harmonic distortion of the D&D below 200Hz is most likely caused by the cabinet design with the side slots and the low crossover frequency of the midwoofer at 100Hz to the rear subwoofers - in case you are wondering about this.
This is why the HD drops a bit below 100Hz when the subwoofers play alone.
right, just wonders was that real distortion vs math error of non conventional design, was wondering so coz during audition it didn't sound so it's real and I will long for the non-cardiod one for some comparison
 
Top Bottom