Modern Measurement Tools Are Tricking Audiophiles Into Trusting Bad Data, Warns Veteran Speaker Designer

[Matt_Holland]

I explained that the RPM in the car doesn't tell you how fast you are going. Then you asked again what the RPM was.

Once again, Klippel NFS uses near-field measurements that are usually in half a meter distance or even less, to compute the far field response of the speaker at any distance. The near-field measurements are internal data and have no relevance. They could be made at 0.2 or 0.6 meter and the outcome would be the same. The reason for variation is the size of the speaker so that the robotic mic arm doesn't hit the speaker. Not that it has any bearing on the final computed frequency response.

If you still don't understand this, then you have no business commenting on what Klippel NFS is capable of, or not.

I’m fascinated by the NFS’s ability to calculate far field response from close mic measurements in the context of the DUT being a large multi-way speaker where layman understanding suggests that correct/intended driver summation doesn’t occur until the measurement position is far enough away, something like 2-3x largest driver spacing. Is there a qualitative explanation you can help with as I highly doubt I could understand the maths (which I found hard enough when taking my acoustics degree 25 years ago!). Thanks.

 

[amirm]

Personally I do not measure loudspeakers. But I am interested in measurement methods. As far as I understand your input the Klippel NFS measures at very near distance the speaker chassis and then computes the sound field at a distance. So you claim the sophisticated mathematic give right results. Is this really true, was this tested?

Of course. I have made a number of comparisons to anechoic measurements and results have excellent correlation -- far better than two anechoic chambers. The mathematics is 100% solid and published. Nothing of what i write is a "claim." The only people posting claims are people who don't understand the system. Start reading here if you doubt the math: https://www.audiosciencereview.com/...nderstanding-how-the-klippel-nfs-works.13139/

I think that this is what AJ said when he meant that automated testing hides what was going on.

That is a claim! I am confident that AJ doesn't understand the mathematics of even the operation of Klippel NFS. This is new technology and requires very advance knowledge of mathematics and acoustics to understand it (see NTK thread above).

 

[SSS]

Of course. I have made a number of comparisons to anechoic measurements and results have excellent correlation -- far better than two anechoic chambers. The mathematics is 100% solid and published. Nothing of what i write is a "claim." The only people posting claims are people who don't understand the system. Start reading here if you doubt the math: https://www.audiosciencereview.com/...nderstanding-how-the-klippel-nfs-works.13139/


That is a claim! I am confident that AJ doesn't understand the mathematics of even the operation of Klippel NFS. This is new technology and requires very advance knowledge of mathematics and acoustics to understand it (see NTK thread above).

Thank You for the information. Instead of claim I should have said statement, sorry tackling you.

 

[Chris A]

I am trying to teach you to not repeat these myths.

By the way, I, too dislike being credited with saying things that I didn't say. So a little reciprocity is requested.

I am confident that AJ doesn't understand the mathematics of even the operation of Klippel NFS.

I see, I was not aware of your belief re: this subject. Thank you for sharing that. I thought all the apparent emotion was directed in my direction.

I'm not sure I can agree or disagree, since I do not know this person--either through reputation or personally.

Chris

 

[amirm]

I’m fascinated by the NFS’s ability to calculate far field response from close mic measurements in the context of the DUT being a large multi-way speaker where layman understanding suggests that correct/intended driver summation doesn’t occur until the measurement position is far enough away, something like 2-3x largest driver spacing. Is there a qualitative explanation you can help with as I highly doubt I could understand the maths (which I found hard enough when taking my acoustics degree 25 years ago!). Thanks.

Sure. If I throw a stone in a pond, you see a bunch of circles expanding. I can model that expansion and using that, predict what the circle looks like at any distance from the stone. The circles would start at the size of the stone and keep getting larger proportional to distance from the stone. This is indeed the behavior of a speaker at very low frequencies (omni directional).

Now if you throw two stones in the water, the same circles expand but now interfere with each other. This again, is simple to model as to expanding waves with different phase.

Now take this to Nth degree of many stones thrown in the water, representing the drivers, cabinet, etc. of the a speaker. A complex sum is now radiating into 3-D space but it can be modeled using simple set of functions. Finding this model is what Klippel NFS does. At low frequencies, its job is simple just like the single stone. At high frequencies, the radiation pattern is very complex requiring combination of "high order" functions.

Key is that once there, you have fully modelled the radiation pattern of anything. The system doesn't care. It captures its radiation at say, 0.2 meter, and then created a model for 1, 5 or 20 meters.

 

[Matt_Holland]

Start reading here if you doubt the math: https://www.audiosciencereview.com/...nderstanding-how-the-klippel-nfs-works.13139/

My previous post was just before you posted this link.
Working my way through it now.

 

[amirm]

I see, I was not aware of your belief re: this subject. Thank you for sharing that. I thought all the apparent emotion was directed in my direction.

I'm not sure I can agree or disagree, since I do not know this person--either through reputation or personally.

Well, I do. The mathematics behind Klippel NFS is impenetrable by any acoustic designer who doesn't have masters degree in signal processing. Here is the formula from NTK thread:

There is no way, no how, a speaker design has any idea what this equation means.

This lack of understanding resulted in another speaker designer claiming that the many measurements Klippel NFS performs, are just averaged to produce the final output! This is of course completely wrong. So when I say people who have not used the system don't understand it, it comes from knowing the topic and layman application of it as addressed by speaker designers.

So that you know that I know what the above means , the green coefficients address how the radiation changes over distance (special category of Bessel functions), and the ones in blue are the basis functions which combine to create the shape of the radiation.

 

[audio_tony]

We know the Klippel NFS is accurate. We have verified many times on ASR by comparing measurements from multiple sources.

I think consistent rather than accurate (not that I'm disputing the accuracy of the Klippel system).

Compared to earlier methods, the Klippel system is probably accurate - but how do we define accurate?

 

[kemmler3D]

************ must have taken on a new social media strategy lately, because I see them popping up with flat out wrong, arguable ragebait all the time lately. I'm not sure if they were always this full of it, but lately I felt compelled to comment to dispute their assertion that vinyl contains more information than high res digital...

I'd be surprised if AJ actually thinks Klippel scans are bad for the consumer... Or himself for that matter, I think Erin has been a big booster of his Mofi designs?

 

[kharan]

@amirm There are two things I'm quite curious about.

When the Klippel software computes the estimated in room response, what's the height of the speaker? Specifically, I mean do the computed response take into account that a bookshelf rests on a stand while a floorstander stands on the floor?

Following that, what's the placement of the speaker and the room like in that computed response? Obviously, the "room" has no nodes, but what kind of parameters does it use for that imaginary room?

 

[Matt_Holland]

@amirm There are two things I'm quite curious about.

When the Klippel software computes the estimated in room response, what's the height of the speaker? Specifically, I mean do the computed response take into account that a bookshelf rests on a stand while a floorstander stands on the floor?

Following that, what's the placement of the speaker and the room like in that computed response? Obviously, the "room" has no nodes, but what kind of parameters does it use for that imaginary room?

I think the key thing is that the NFS calculates the free field response. Therefore no boundary interference at all. We want this consistent approach for all speaker measurements otherwise we can’t make correct comparisons between speakers.

 

[kemmler3D]

I think the key thing is that the NFS calculates the free field response. Therefore no boundary interference at all. We want this consistent approach for all speaker measurements otherwise we can’t make correct comparisons between speakers.

I think the question is about the stimulated in room response generated from the free field response, namely what is this stimulated room like, specifically? I understand it's meant to be a typical generic sort of room but I agree it would be interesting to get more details.

 

[Arindal]

It assumes it is an arbitrary source radiating sound.

That is exactly the problem with NFS in my understanding, that it tries via iteration, to calculate a number of imaginary sound sources at unknown positions, which would fit to the soundfield it has actually measured at numerous positions in what is neither nearfield nor farfield. If the actual number of sound sources is closer to indefinite, they are far away from each other compared to the wavelength, or their phase relations towards each other are chaotic, that cannot be as accurate as claimed, as some measurements with bending-wave planar transducers or cardioids have proven.

And yes, I am understanding how the NFS calculation works, and that it would in theory be perfectly working for any planar diaphragm moving in a pistonic manner with no major chaotic cancellation effects at play.

I have yet to see an anechoic chamber that is flat down to 20 Hz which Klippel NFS is. You also need an awfully large anechoic chamber to be able to measure at 3 meters.

With a very huge room, you can go pretty low just with the help of time windowing. Have you been to the cube in Struer?

Btw you do not need to go flat as low as 20Hz. Klippel does not do that accurately as well. Otherwise you would never get a reading of negative directivity index with speakers that solely employ omnidirectional bass sources without major phase differences. Not a big deal, though, as the frequency plot itself is not too far off in most of standard cases, but indirect indication of limitations of the calculation. It simply should not happen.

Now take this to Nth degree of many stones thrown in the water, representing the drivers, cabinet, etc. of the a speaker. A complex sum is now radiating into 3-D space but it can be modeled using simple set of functions. Finding this model is what Klippel NFS does.

It is a good analogy. Now imagine an indefinite number of stones being thrown, slightly delayed in a chaotic manner, plus stones being thrown from under the water hitting the surface of the water, or little pumps sucking it in. That is basically what a bending-wave transducer like a huge planar does. In my understanding, even the most sophisticated model comes to its limits trying to calculate an accurate representation of that chaos. You have indirect indication for this, the moment the error rate increases.

the Klippel NFS measures at very near distance the speaker chassis and then computes the sound field at a distance. So you claim the sophisticated mathematic give right results.

I have the advantage of having worked with truly anechoic measurements of existing loudspeakers for decades, manyfold the number of what Amir has measured, as well as with Klippel systems, so I had the chance to compare.

Can tell you that if the aforementioned conditions for potential miscalculations, such as chaotic out-of-phase behavior of parts of the diaphragm, huge distances or cancellation effects between sound sources (as typically found with large planar transducers, cardioids, dipoles, line sources and alike), are NOT in play, and the wavelengths are not overly long (i.e. everything above 100Hz for a room of normal size), the NFS calculation is very very accurate. I have no reasons to doubt the results except for very unusual cases.

 

[Chris A]

The mathematics behind Klippel NFS is impenetrable by any acoustic designer who doesn't have masters degree in signal processing.

This subject is actually my second experience with the nearfield/farfield problem. The first was in nearfield vs. farfield measurement of Vibroseis seismic sources for land geophysical exploration. That domain is complicated by the fact that other modes are present in solids (the earth) rather than gasses (air). The speed of sound in rock formations is also much faster than air, so the nearfield extends to a very large radius at low frequencies. Non-homogeneity of the transmission medium is also a significant issue. Corrections to nearfield measurements do have limitations, even if the limitations are inconvenient to acknowledge.

Also, assumptions on people's understanding of physical phenomena and even signal processing algorithms isn't always evident from simply looking at CVs. I've learned to at least initially extend the benefit of the doubt.

Chris

 

[Matt_Holland]

I think the question is about the stimulated in room response generated from the free field response, namely what is this stimulated room like, specifically? I understand it's meant to be a typical generic sort of room but I agree it would be interesting to get more details.

Ah I see. Thanks for clarifying.
I would be interested in understanding this too.

 

[kharan]

I think the key thing is that the NFS calculates the free field response. Therefore no boundary interference at all. We want this consistent approach for all speaker measurements otherwise we can’t make correct comparisons between speakers.

What kemmler3d said. And also, the floor and ceiling reflections would be impacted according to the height of the speaker.

 

[Sokel]

I think the question is about the stimulated in room response generated from the free field response, namely what is this stimulated room like, specifically? I understand it's meant to be a typical generic sort of room but I agree it would be interesting to get more details.

Not only the room but its relation to the speakers, listening distance, etc.
Such a resulting curve can't be at the meter for example.

 

[DanielT]

Comment/opinion from Andrew Jones UK loudspeaker designer. I think he has some valid points to consider in loudspeaker measuring.

This site is not one I follow closely I rarely use headphones.

Robert

Andrew Jones tries to satisfy both objectivists and subjectivists. To subjectivists by throwing in a few bones to them that measurements is difficult.
That's what they want to hear so they can then INTERPRET it as: Measurements can't be trusted, they're not good, the fact is they don't say anything, it's my golden ears that are the yardstick.

That objectivists, with commercial intentions, satisfy the potential subjectivist large buying group we see all the time. It's from how
Bruno Putzeys in interviews talks about how his class D creations sound like to how Sterophiles' own John Atkinson with masterly diplomatic and vague language speaks about how crappy measuring and damn expensive (a not too unusual combination) are. Almost a kind of own language then: Atkinsonian
What don't you do to keep advertisers in a good mood.

 

[Arindal]

Andrew Jones tries to satisfy both objectivists and subjectivists. To subjectivists by throwing in a few bones to them that measurements is difficult.

I don't think so. My experience with AJ is, he is both an objectivist and subjectivist at the same time and genuinely trusts his listening test results as much as his measurements, believing that both camps should take advantage from the other one´s findings.

Could not agree more with him. Hearing his TAD R1 almost two decades ago, was the decisive moment for me to realize what really differentiates a great sounding speaker from one which just does avoid the common mistakes.

 

[Curvature]

I think consistent rather than accurate (not that I'm disputing the accuracy of the Klippel system).

Compared to earlier methods, the Klippel system is probably accurate - but how do we define accurate?

Here's one example: https://www.audiosciencereview.com/...n-axis-measurements.20050/page-3#post-2330133

Lots of other examples that can be viewed more easily on spinorama.org by @pierre. The best is probably the KH80, which was measured by @amirm three times using multiple units, and also measured by Erin, @napilopez, @Nuyes, and Anselm Goertz at Sound & Recording. The latter two sets of data aren't featured on spinorama.org, however. Regardless, they represent a diversity of techniques and the measurements in some cases were taken years apart. Despite all that they align well with Neumann's own comprehensive measured data, and Neumann themselves reviewed @amirm's work.

Erin, in his early reviews, compared his manual work (speakers up on a ladder) to groundplane measurements, and did the same in a few cases even when he bought the NFS.

The main issues affecting physical accuracy of the NFS are its vibrations being commingled with speaker results and the accuracy of the microphone. I think that accounts for the difference between those results and those from anechoic chambers, which are typically only rated as such to 100-200Hz, below which a compensation curve is applied to measurements. There differences are in trends within 1-2dB, sometimes larger with other lower resolution techniques, and without doing a lot of work that's probably the best we can hope for.

For example, what would it take for the NFS to deliver physically true results within 0.1dB? We would need a completely known reference and a number of careful compensations in a completely controlled environment. That's not practical at all given what we know and the tools available. It's also too much to ask of Amir, Erin or any other reviewer working outside of a scientific or academic context without funding, a facility and a team of supporting people.

Certainly none of this invalidates measurements as a way of making purchasing decisions.

I don't even know how much of the discussion on accuracy even matters when the bigger issue is that few brands even offer measurements of their own (edit: which are supposedly more accurate than third-party results, NFS or not).