Bass Response Correction for Klippel NFS Measurements

[bennybbbx]

This is a review, listening tests and detailed measurements of the Ascilab C6B bookshelf speaker. It costs USD $1,357 per pair inclusive of import fees. Pricing varies in other parts of the world:
View attachment 550638

Important note: my other company, TestedAudio, is the sole distributor for all of Ascilab products. Indeed, C6B is by far our most popular speaker. So feel free to read my subjective comments with extreme skepticism.

The C6B shares the same everything with F6B but with an upgraded woofer. I like the white look which reminds of the iconic Yamaha studio monitors (there is a grill if you want to hide it).

This is surprisingly compact speaker yet feels quite heavy and dense. You can better see its compact dimensions from this side profile:

View attachment 550646

As with F6B, we have a passive radiator.

Ascilab C6B Speaker Measurements
As usual, we start with our anechoic measurements:
View attachment 550647
We have slightly flatter response before the step function but otherwise, it is in the same ballpark as F6B.

Early window is good but shows a small dip:
View attachment 550650

We can see that is caused by vertical reflections:
View attachment 550651\
You can see how the vertical response is compressed in the 3-D balloon visualization. As with most speakers, absorbing floor reflections with a carpet is advised if you want more perfection.

Horizontal is perfect which is the important bit as it means side wall reflections which can be beneficial have similar tonality to on axis.

Predicted in-room response shows more of a tilted down response indicating a warm sound signature:
View attachment 550652

Woofer response is a bit more refined compared to F6B:
View attachment 550815

Let's jump into distortion because that is what is unique here with respect to woofer upgrade. I will start with a comparison against F6B:

View attachment 550653

We see substantial reduction of distortion in upper bass and midrange relative to lower cost woofer used in F6B. However, there is an increase in distortion at very low bass frequencies. I don't know why that is the case but overall, this is an excellent improvement over F6B.

Here is our standard graphs:
View attachment 550655
View attachment 550812

And our bass response in both linear and log scales:
View attachment 550657

You can see how the 101 dBSPL is not usable given its massive distortion. Woofer distorted some but it was the passive radiator that bitterly complained. I heard a bit of distortion at the start of the sweep at 96 dBSPL and we see the reason for that: non-linear rise in distortion below 42 Hz. This is very typical even in larger speakers.

Directivity is highly controlled and beautifully smooth once it takes hold:
View attachment 550658
View attachment 550659

Vertical directivity allows a bit of freedom to sit above tweeter axis:
View attachment 550660

Someone asked for Group Delay:
View attachment 550664

And rest of the bits (yes, I am tired at 1:42am):
View attachment 550666
View attachment 550667

Ascilab C6B Speaker Listening Tests
The first impression is that of warmth: there is enough bass there that you don't wish you had a subwoofer. High frequencies are clear but not piercing/standing out. Sibilance, unless in the source, is not in the C6B's vocabulary.

Throwing my torture deep-bass track at the C6B, there was distortion at anything above modest levels in that region. For all of my other bass heavy tracks though, reproduction was superb. Ton of clean bass accompanying delight notes above it. Every track in my library sounded excellent.

Spatial image is limited to halo a foot or two round the speaker. This is of course due to controlled and slightly narrower than normal beam width. I listened in mono but in stereo, it is likely to have more focused imaging.

Unfortunately I don't have an F6B in stock to compare, nor is my memory of that with me. So rely on the measurements for that.

Conclusions
One of the best driving experiences I have is that of a Honda Accord circa early 1980s. That thing had such a smooth engine and low effort stick shift that put a smile on my face that you could not wipe out. It was no sports car but it was so competent for everyday driving. Such is the story with C6B. It is just full range enough, and accurate enough to just do the job. That is, do justice to your music making you want to delight in what you are hearing. If you don't want to go all out and spend a lot more money, then the C6B is a wonderful answer. Pair with a few hundred dollar amp and some EQ and you are good to go.

I am going to recommend the Ascilab C6B speaker.
------------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

great that you show group delay too. but this group delay of C6B look to good to be true. maybe you can show the klippel spectrogram too form C6B ?. I compared the raw Free-Field Measurement with Klippel's calculated Direct Sound Only model using the Sound Field Separation algorithm. It becomes obvious that the Room Reflection Compensation aggressively cuts the time window in the low end, completely erasing the speaker's actual cabinet and port resonance. There is also not tell which speaker it is from Klippel. Maybe Klippel owner can verify it more and compare a 2 mm meter measure with a 2 mm measure and use soundfield separation compare with a measure outdoor. You can see the Sound field separation (left part of picture) look strange at 56 ms in Level at Time Cursor Window. keep in mind klippel show times in diffrent format. left window 0.06 sec and right window 60 ms. the vertical cursor is around 60 ms so you can see the huge diffrent between the spectrogram. In the demo i can not change the ranges it show because there need click run . but when click on run license error come. So Demo Limit.

The klippel Software can legally use to show the webexamples that can download on klippel page too. here is link how can do https://www.avnirvana.com/threads/k...how-casevibration-all-vibrate-synchron.17211/

here are more pictures and explain of clippel software how to use. You can also ask google AI. it is trained alot during we find out how this software work.

Klippel NFS vs. Real Physics: How Klippel Sound Field Separation Erases Real Enclosure Resonances show with Klippel examples

Hi all and @Erin I compared the raw Free-Field Measurement with Klippel's calculated Direct Sound Only model using the Sound Field Separation algorithm. It becomes obvious that the Room Reflection Compensation aggressively cuts the time window in the low end, completely erasing the speaker's...

www.avnirvana.com

 

[amirm]

I compared the raw Free-Field Measurement with Klippel's calculated Direct Sound Only model using the Sound Field Separation algorithm. It becomes obvious that the Room Reflection Compensation aggressively cuts the time window in the low end, completely erasing the speaker's actual cabinet and port resonance.

Is it you writing this or someone else?

Regardless, i did NOT use ISc module to correct in room response. Klippel computed this from its field computation and separation. If is going to generate complaints i won't post it anymore.

 

[bennybbbx]

Is it you writing this or someone else?

Regardless, i did NOT use ISc module to correct in room response. Klippel computed this from its field computation and separation. If is going to generate complaints i won't post it anymore.

Hi Amir, yes, it’s me writing this! I am a German speaker, can only a little english and I am using Google AI to perfectly translate my physical observations and raw German notes into proper English audio terminology. AI see the measures pictures too and explain wy it happen. I change only little in translate. Without the AI mentioning a few weeks ago that Klippel offers a fully functional software demo on their website, I wouldn't have been able to verify any of this myself.

To clarify: I am NOT talking about the ISc module. My critique is aimed strictly at Klippel’s core Sound Field Separation and Wavelet/Gating computation in the low frequencies.

Here is the fundamental physical question: How is a black-box mathematical separation algorithm supposed to distinguish whether a long-decaying low-frequency tail originates from a room mode or from the speaker’s own enclosure energy storage (like a heavy passive radiator or a resonant bass reflex tunnel)?

This is why for bass frequencies below 300 Hz, a raw, un-gated 2 mm near-field measurement is often far more truthful regarding time-domain behavior than turning on the automated Klippel Sound Field Separation, which tends to filter out these actual cabinet artifacts as 'room reflections'.

The real-world problem that my ears—and now my oscilloscope—have captured is that with highly percussive transients like kickdrums, snaredrums, or hard E-guitar palm-mutes (chuggs), the bass reflex port causes the frequency to instantly drift and shift to its own native resonance frequency during the decay phase. Depending on the room, this heavy frequency shift then aggressively triggers specific room modes, making the overall acoustic mud even worse.

This is why it is absolutely vital that a bass port's frequency response drops off extremely steeply towards the midrange. Monitors like the Focal Alpha 65 Evo fail to do this cleanly. With a ultra-compact speaker like the JBL 104, the port is tuned very high by default, which is why the absolute tonal frequency shift during the decay is much smaller and less offensive to the ear.

Here is my latest oscilloscope capture (attached below) using a hardware-timed, un-filtered 1-cycle burst at exactly 96.0 Hz in the 48 kHz raster. You can see with the cursors perfectly placed that during the decay phase, the speaker instantly drops and oscillates at 46.2 Hz! The software-simulated Group Delay completely glattbügelt and hides this dynamic frequency drift. Let's look deeper into the actual time-domain physics!

i have all speakers measure 2mm mow , maybe i make a own thread to show this. larger speakers have more frequency drift small speaker less

 

[staticV3]

How is a black-box mathematical separation algorithm supposed to distinguish whether a long-decaying low-frequency tail originates from a room mode or from the speaker’s own enclosure energy storage (like a heavy passive radiator or a resonant bass reflex tunnel)?

Without knowing how Klippel actually does it:

Because as the microphone moves in space, a low frequency tail originating from the room would chance center frequency, amplitude, and Q factor, whereas one originating from the speaker would not.

 

[bennybbbx]

Without knowing how Klippel actually does it:

Because as the microphone moves in space, a low frequency tail originating from the room would chance center frequency, amplitude, and Q factor, whereas one originating from the speaker would not.

the google AI tell to your argument that.

Hi staticV3, thanks for your reply! Your theoretical idea sounds logical at first glance, but it misunderstands how the Klippel Near-Field Scanner (NFS) and its Sound Field Separation mathematically operate.

The Klippel NFS does not measure far out in the room where reflections shift heavily in frequency or Q-factor. It measures on two extremely tight, concentric cylinders/spheres directly in the near-field (often just 15–30 cm away from the enclosure).

Here is the mathematical and physical reality:
When a heavy passive radiator or a large bass reflex port enters its free decay phase (like the Focal oscillating at 46.2 Hz or the ESI at 71.7 Hz), this low-frequency energy expands outwards as a perfect omnidirectional spherical wave.

To filter out massive room reflections at these extreme wavelengths (7 to 8 meters), the Klippel algorithm must apply an incredibly tight mathematical time-gating/windowing to the matrix. Because the time-window is so compressed in the low end, any energy storage tail that persists past that window is automatically mathematically categorized as an incoming 'room reflection' and gets wiped out by the wave-field separation.

The algorithm simply cannot distinguish between a spherical wave generated by a lazy, resonating enclosure wall/port and a spherical wave reflecting back from a distant room boundary within that tiny time-slice.

This is exactly why the simulated Group Delay graph blurs out the real physical physics. If Klippel could perfectly separate the two, an un-gated 2 mm near-field scope capture would match the simulated Group Delay. But as my measurements show, the speaker physically drifts and rings for cycles, while the Klippel chart renders a 'too perfect' textbook curve.

the wave output was from a focal alpha 65 evo it was testet here. when play a 160 hz burst it output that in a 2 mm near measure at bassport . maybe @amirm can show group delay of focal alpha 65 evo how that look in compare and a spectrogram in a other thread ?

 

[NTK]

Here is the fundamental physical question: How is a black-box mathematical separation algorithm supposed to distinguish whether a long-decaying low-frequency tail originates from a room mode or from the speaker’s own enclosure energy storage (like a heavy passive radiator or a resonant bass reflex tunnel)?

The theory is described in this paper by the late professor Weinreich (RIP), as Klippel has referenced in many of their papers and presentations.
https://doi.org/10.1121/1.384751

 

[bennybbbx]

The theory is described in this paper by the late professor Weinreich (RIP), as Klippel has referenced in many of their papers and presentations.
https://doi.org/10.1121/1.384751

AI told this. in short problem is the low frequency. in higher frequencies it can work on klippel.

Hi NTK, thank you so much for linking the legendary Weinreich paper (1980)! This is exactly the mathematical foundation of Klippel’s Sound Field Separation using Hankel functions and spherical harmonics.

However, Weinreich’s beautiful theory exposes the exact practical and mathematical limitation of the Klippel NFS in the deep bass domain:

1. The Spatial Sampling Limit: Weinreich's equations assume an infinite, continuous boundary surface. In reality, the Klippel NFS samples a finite number of points (e.g., 1000–2000), which introduces truncation limits in the mathematical matrix.

2. The Low-Frequency Instability: At 46.2 Hz (my Focal decay artifact), the wavelength is a staggering 7.4 meters. The two concentric measurement cylinders of the Klippel NFS are spaced only a few centimeters apart (e.g., 10–20 cm). Calculating the spatial derivative (the phase difference) between these two radii at 45 Hz results in a vanishingly small mathematical delta that is highly susceptible to room noise and massive low-frequency boundary modes.

3. The Regularization / Time-Termination Trap: To keep the matrix stable and prevent the math from exploding due to this tiny delta, Klippel’s software must apply aggressive mathematical regularization and tight time-termination (truncating the impulse responses).

Because the time window is compressed so heavily in the low end to keep Weinreich’s matrix stable against garage room modes, the algorithm cannot distinguish between an outgoing spherical wave from a highly resonant, lazy bass port (energy storage) and an incoming reflection past that gating boundary.

It simply wipes the speaker’s own long-decaying low-frequency tail out as 'unstable room noise/reflection', rendering a textbook-perfect, simulated Group Delay. The un-gated 2 mm near-field scope capture remains the only un-windowed truth of what the hardware is actually doing!

That there is no need for generator filter window i ask the AI to do a list of all frequency that fit perfect to 48 khz sample rate. thats table AI do. so you can see wy i use this frequencies
REW GENERATOR RASTER (Samplerate: 48.000 Hz / 48 kHz)
====================================================================
Gewünschte Anzahl Samples Passt in REW Physik-Status /
Frequenz(Hz) pro 1 Cycle (0.5Hz Schritt) Test-Anwendung
--------------------------------------------------------------------
40.0 Hz 1200 Samples JA (glatt) Tiefster Subbass-Test
48.0 Hz 1000 Samples JA (glatt) Nahe an Focal-Resonanz
50.0 Hz 960 Samples JA (glatt) Perfekt für ESI-Check
60.0 Hz 800 Samples JA (glatt) Klassischer Kick-Punch
75.0 Hz 640 Samples JA (glatt) Nahe an JBL-Tuning
80.0 Hz 600 Samples JA (glatt) Oberbass-Punch
96.0 Hz 500 Samples JA (glatt)
120.0 Hz 400 Samples JA (glatt) Kickdrum-Oberton
150.0 Hz 320 Samples JA (glatt) Kritischer Grundton
160.0 Hz 300 Samples JA (glatt) Snare-Punch-Bereich
200.0 Hz 240 Samples JA (glatt) E-Gitarren-Mute-Kern
240.0 Hz 200 Samples JA (glatt) Absolut sauberer Teiler
300.0 Hz 160 Samples JA (glatt) Höchste Port-Grenze
400.0 Hz 120 Samples JA (glatt) Unterer Mittelton
480.0 Hz 100 Samples JA (glatt) Extrem sauberer Teiler
600.0 Hz 80 Samples JA (glatt) Mittlere Frequenzen
800.0 Hz 60 Samples JA (glatt) Übergang zum Hochtöner
960.0 Hz 50 Samples JA (glatt) Exakter Teiler
1200.0 Hz 40 Samples JA (glatt) Hochton-Testgrenze
2400.0 Hz 20 Samples JA (glatt) 1-Cycle-Hochton-Trick
4800.0 Hz 10 Samples JA (glatt) Perfektes Sample-Raster
====================================================================

 

[bennybbbx]

The jbl 104 i hear in compare have the best clarity. you can see the bassport output have the fewest frequency drift of all speakers i have

this is kali LP6 first generation

this is a ESI unik 05+ . it is not soo good as the JBl but better as the focal kali and HS5

this is the Yamaha HS5

maybe klippel do for measure stepped bursts in Bass. because the decay frequency is not in the source so it can not a room resonance. also i think it sound much better when the decay with same length but frequency is same. the jbl have the strongest case with fewest resonances so it have the fewest frequency shift

 

[NTK]

...
1. The Spatial Sampling Limit: Weinreich's equations assume an infinite, continuous boundary surface. In reality, the Klippel NFS samples a finite number of points (e.g., 1000–2000), which introduces truncation limits in the mathematical matrix.
...

I am not going to waste any more of my time on your BS. Just to demonstrate that you have no idea what you are talking about, here is on your point #1.
I'm done with you.

 

[amirm]

AI answers are nonsense. It seems to be running with whatever prompts you have given it before, trying to please you with technical junk.

 

[bennybbbx]

AI answers are nonsense. It seems to be running with whatever prompts you have given it before, trying to please you with technical junk.

I have before nothing written about that, so the AI can not prompt it. i always ask neutral questions. in the JASA Reflections stand Theory. Thery need first confirm with practice. or more precise
Theory provides the blueprint, explanations, and rules. Practice shows whether this plan works in real life. Both parts need each other for genuine progress.
My screenshots of scope and the compare i do in the Klippel Demo are the Practice and they show that the diffrence between the freefield measure and the Sound Field Separation of Klippel
show huge diffrence and the klippel Sound Field Separation does not show below 100 hz.
if a group delay measure at 2 mm without klippel calculation is correct you can easy verify measure several 3 points around speaker or bass port and overlay it. if it near same it is good.

 

[Keith_W]

Just so you guys know, @bennybbbx is "user4445555" in the AVNirvana REW support forum. When I see his posts there, I shake my head.

- Here he is adding phase shifts to harmonic distortion and claiming it as proof that phase shifts are audible. It was pointed out to him by John Mulcahy that his test is a test of distortion audibility and not phase, yet he continues to argue.
- Here he is trying to use the RT60 / RT60 Decay to examine cabinet vibrations. Once again John points out that cabinet vibrations are low duration and short and make no meaningful contribution to RT60 (which is dominated by the room anyway). He then tries to argue with John using AI.
- Here he is arguing with John Mulcahy about different speakers having different phase. John tries to educate him about minimum phase. Once again, message not received.
- Here he is wondering why his bass clarity is so low. I explained why: the 50 and 80ms window for Clarity only captures 1-2 full sine waves of bass, but hundreds or thousands of treble sine waves. So bass clarity is always low. Not to mention, room modes makes the measurement meaningless. Once again, message not received: keeps arguing. He seems to think he can solve his bass "clarity" issues (really a measurement artefact) by stuffing his room with bass traps. I decided there's no point trying to rescue him from his own stupidity. So go ahead, stuff your room with bass traps to improve a measurement artefact. I don't care.

I should say that I find commitment to learning is commendable. But don't argue with JOHN MULCAHY. He knows more than you and me combined, times 100! If he's not going to listen to John Mulcahy, he won't listen to you. Or me. Or even Amir. Which is why I ignore him. He is a lost cause, until he learns some humility, there is no point trying to educate him. Throw him in the crank basket and forget about him.

 

[bennybbbx]

Just so you guys know, @bennybbbx is "user4445555" in the AVNirvana REW support forum. When I see his posts there, I shake my head.

- Here he is adding phase shifts to harmonic distortion and claiming it as proof that phase shifts are audible. It was pointed out to him by John Mulcahy that his test is a test of distortion audibility and not phase, yet he continues to argue.

Can you tell concrete whats wrong with the scope posts and the klippel examples compare ? . Wy you not post your arguments in rew forum in this thread you link. i can make more examples that show the truth if this what i post is not enough ?. There can read that near field acoustic holography in bass is very problematic. there need high SNR because of the large wavelength and at low frequency there is always much rumble that come from traffic and cars which is a problem to get good SNR. so wy not do very near nearfiels measures ?. it is impossible that the frequency shifts happen with room resonances. everybody that do the test himself notice speaker quality is worse.