Understanding How the Klippel NFS Works

[MZKM]

Here is what I mean about this stuff taking a long time and it being damn near impossible to do these lengthy tests

Yeah, yeah, give us cabinet vibration tests old man.

Joking of course. But can that data be extrapolated from any currently done measurement (extreme would result in visible resonances)? Because if the speaker isn’t well damped and the cabinet vibrates like heck during bass notes, that is of course not ideal. Wouldn’t cabinet vibrations mean a loss in acoustic transfer and thus SPL is lower?

If not, a “I put my hand around the cabinet while playing bass and it felt inert” would work if you can fit that in during your subjective listening; if not, of course that’s totally fine.

Also, circling back to your video with the guy from Klippel when you talked about phase: I was wondering why impedance/phase plots show resonances, and while I’m not too sure on impedance, one explanation I found online is that phase is related to the derivative of the FR, and thus when the FR suddenly spikes the phase will too (and also that group delay is a derivative of phase shifts that aren’t minimum phase).

 

[hardisj]

If not, a “I put my hand around the cabinet while playing bass and it felt inert” would work if you can fit that in during your subjective listening; if not, of course that’s totally fine.

I don't understand why people even do that silliness. Or the accelerometer tests (I've done it before; I stopped).

Resonances show up in an impedance plot and the FR. I don't think I've ever needed anything other than one of those.

 

[MZKM]

I don't understand why people even do that silliness. Or the accelerometer tests (I've done it before; I stopped).

Resonances show up in an impedance plot and the FR. I don't think I've ever needed anything other than one of those.

Ok, didn’t know if only extreme cabinet vibrations showed up (but then, that’s all you really care about anyway).

 

[hardisj]

Ok, didn’t know if only extreme cabinet vibrations showed up (but then, that’s all you really care about anyway).

A really good example is of the Klipsch Heresy IV that I reviewed. I heard strong resonance in the lower male vocals. I made a note that it sounded somewhere between 100-150Hz. I took it to an entirely different room and still heard the issue. When I did the measurements the impedance sweep showed a blip at around 110 Hz. That was the problem.

 

[dc655321]

Also, circling back to your video with the guy from Klippel when you talked about phase: I was wondering why impedance/phase plots show resonances, and while I’m not too sure on impedance, one explanation I found online is that phase is related to the derivative of the FR, and thus when the FR suddenly spikes the phase will too (and also that group delay is a derivative of phase shifts that aren’t minimum phase).

Chapter 3.6 "Anomalous Impedance Data" in "Testing Loudspeakers" (J. D'Appolito) has an approachable description of resonance effects manifesting as electrical impedance effects.

Edit: some helpful, practical examples in that section too.

Essentially, changes in mechanical/acoustic driver loading will appear as electrical effects. Not surprising given that loudspeakers are electric-acoustic transducers, and must exchange energy between those domains...

 

[Few]

Any hope that the holographic Klippel measurements can point to significant enclosure radiation sources? They’re probably down too many dB, but it would be fun to see a little blip in a bubble plot (or some other) that could be tracked down to a panel being the sound source.

 

[MZKM]

Can't believe I didn't think to post this here. I had Christian Bellmann (one of the inventors of the NFS) on my channel to discuss the NFS. You guys may find this useful.

Just made this post in another thread, but I’ll post it here too and ask you the same question (last paragraph):

Before I show you the directivity plots, let me post a new measurement I have not shown before which indicates at what distance the speaker acts as if it is in far field:

This says that above 400 Hz, that distance is 1.5 meters (where the circles is on blue line) Lower frequencies take forever to get this way so I have excluded them.

Not sure if you watched Erin’s (@hardisj, just to tag him) talk with Klippel (1hr 23min mark):

You don’t look at the highest order to determine far-field, you look at Total Power. Look at how huge your y-axis is, that is throwing off the intuitive nature, we don’t need the 13th spherical harmonic to be -200dB down from the monopole. He states (as well as one of Klippel’s PDFs) that the far-field limit is when Total Power reaches 0.5dB from being monopole.

Also, showing it at 400Hz just shows it at 400Hz, not it and above. So at 400Hz it’s not 1.54m but around 30cm (your scale is too huge to see 0.5dB increments). This makes sense as it’s mostly the woofer playing that frequency. You stated as you went lower in Hz that the distance increased, which makes sense as this is a rear-ported speaker so the sound field is more complex in the bass and you need to be further away for the port and woofer to sum.

The guy at Klippel said that they are working on generating far-field transition distance for all frequencies, but that it needs work.

However, due to my understanding, there is a manual way. You first look at the Radiated Sound Power graph:

And look at where the monopole nature is most reduced, in this case around 1500Hz.
I am willing to bet if you look at the Apparent Sound Power for any frequency for this speaker above that ~1500Hz will have the furthest distance to be considered far-field.

_______
Assuming the Apparent Sound Power data can be exported, can it only be exported as 1 frequency (what the graph is limited to), or can you export it for all frequencies? Because if so and you are willing, I can find the far-field distance.

 

[MZKM]

Any hope that the holographic Klippel measurements can point to significant enclosure radiation sources? They’re probably down too many dB, but it would be fun to see a little blip in a bubble plot (or some other) that could be tracked down to a panel being the sound source.

The Directivity ballon plots will tell you this. But you would have to go frequency by frequency to see this.

 

[Few]

The Directivity ballon plots will tell you this. But you would have to go frequency by frequency to see this.

Got it. Thanks. I guess you could just focus on frequencies that look suspicious on impedance or frequency response measurements.

 

[MZKM]

Got it. Thanks. I guess you could just focus on frequencies that look suspicious on impedance or frequency response measurements.

Yeah, I don’t remember what speaker, but in one of Amir’s reviews he shows the balloon plot and you can see the cabinet diffractions.

 

[NTK]

Any hope that the holographic Klippel measurements can point to significant enclosure radiation sources? They’re probably down too many dB, but it would be fun to see a little blip in a bubble plot (or some other) that could be tracked down to a panel being the sound source.

The more typical application of nearfield acoustic holography (NAH) is to identify the vibration mode shapes of the vibrating structure. Therefore, the method should be able to reveal structural resonances of the speaker cabinet. This is, after all, the NAH method's claim to fame.

The Klippel NFS, as it is currently implemented, is not suitable for this application. To measure the structural vibration mode shapes, the measurements need to be taken very close (a few millimeters) and be conformal to the structure. Below is a picture I found on the web that show how the structural vibration of a guitar was measured.

Measuring the vibration mode of the structure (using NAH) is what is called an inverse problem. We measure the sound pressure field some distance away from the structure, and calculate backward (with respect to the traveling direction of the sound waves) to figure out what happened at the source surface. The Klippel NFS, on the other hand, solves the forward problem -- predicts what will happen further away from the measurement points. The forward problem is the easier problem. What happens further away is fully determined by what passes through the measurement zone. The inverse problem is more difficult as not every vibration wave propagates -- some attenuate very quickly (decrease exponentially with distance) and are imperceptible unless at very close distances.

However, the actual vibration modes of the structure is really of interest only to the speaker designers. Everything audible is captured by the NFS. We may not know exactly how it came about, but the effects, if audible, are measurable by the NFS. For the speaker designers, they can analyze the vibration modes, see which ones generate audible resonances and which ones don't, and target structural reinforcements and/or damping to only eliminate the ones that are audibly problematic.

Theories aside, the polar directivity plots from the NFS are interesting. Even though the speaker is supposed to be non-directional at the very low frequencies, the polar directivity plots of the Revel M105 showed they originated from the rear port.

 

[bennybbbx]

which rise time do the waterfall mresures use ?. REW do as default a very large rise time. but rise time should be short to detect if a speaker is slow or not. stereophile use 0.15 ms. can this in klippel use too ?

 

[NTK]

There is no "waterfall" measurement, AFAIK. All measurements are done with logarithmic sine sweeps. From these sweeps, the free field complex frequency response is reconstructed. Using the assumption/approximation that the system is LTI (otherwise it would be rather pointless to talk about frequency response), the impulse response, step response, waterfall, etc. can be generated.

 

[mwmkravchenko]

Chapter 3.6 "Anomalous Impedance Data" in "Testing Loudspeakers" (J. D'Appolito) has an approachable description of resonance effects manifesting as electrical impedance effects.

Edit: some helpful, practical examples in that section too.

Essentially, changes in mechanical/acoustic driver loading will appear as electrical effects. Not surprising given that loudspeakers are electric-acoustic transducers, and must exchange energy between those domains...

Generally you want to have these three on a graph SPL, Phase and Impedance. When there is a lining up of bumps or even a little lag in the phase but close to the same frequency you have something to investigate. Loudspeaker enclosures are a lot more complex than what is generally thought. And a great deal of them are built for economy not real top performance. They get treated like a black box that is only a temple for the driver gods. Taint how it actually works. The box can make a or break even a stellar performing driver. And a really well worked out enclosure can make a decent driver shine very brightly.

 

[bennybbbx]

There is no "waterfall" measurement, AFAIK. All measurements are done with logarithmic sine sweeps. From these sweeps, the free field complex frequency response is reconstructed. Using the assumption/approximation that the system is LTI (otherwise it would be rather pointless to talk about frequency response), the impulse response, step response, waterfall, etc. can be generated.

I know that this can do from measurement. Have the Klippel no rise setting ?. here is from REW with diffrent rise settings. at 10 khz period time is 0.1 ms. and when the waterfall show a time more than 1 ms over more frequency around 10 khz this can not correct, because this mean the speaker do decay always in 10 cycles. here can see a waterfall with diffrent rise times. you see the decay get faster when use shorter rise time. maybe in the klippel waterfall view can reduce this time ?

 

[NTK]

I know that this can do from measurement. Have the Klippel no rise setting ?. here is from REW with diffrent rise settings. at 10 khz period time is 0.1 ms. and when the waterfall show a time more than 1 ms over more frequency around 10 khz this can not correct, because this mean the speaker do decay always in 10 cycles. here can see a waterfall with diffrent rise times. you see the decay get faster when use shorter rise time. maybe in the klippel waterfall view can reduce this time ?

I have never used the Klippel software and don't know what available user adjustable parameters are in its waterfall or CSD display settings.

Sorry.

 

[amirm]

I know that this can do from measurement. Have the Klippel no rise setting ?.

It does. I use 0.5 millisecond.

 

[bennybbbx]

It does. I use 0.5 millisecond.

maybe this can set to lower value ?
here are waterfall with 0.5 0.2 and 0.1 ms. 0.15 ms (as stereophile use)is not possible in REW. but 0.2 ms look good too. I use for all screenshots 100 segments so you can count the segment lines at 7 khz until it reach 65 db. at 0.5 ms rise time i ccount more than 12 segment lines .at 0.2 ms rise time i count 6 segmentlines. at 0.1 ms risetime 5 segment lines. you can see that the decay time look much longer at 0,5 ms rise time.

 

[amirm]

I can set it to anything. You haven't explained why I should.

 

[bennybbbx]

My english is not so good. the time that the waterfall show is 2,94 milliseconds and the waterfall contain 100 segments. each segment(slice) have a line. each line is time 2.94 ms /100 0.0294 ms .with the rise time 0.5 ms there are 4 slices see at near same level at 20 khz. and 4 slices are time 0.0294*4= 0.1176 ms. but 20 khz have a period time of 0.05 ms. so i think when the decay time is so slow as the waterfall show this speaker can not play 20 khz. with rise time of 0.2 or 0.1 sec it look more as reality because when look at the first 4 slices after the impulse the decay time is much more