Harmonics being stripped & Measurements not adding up. SMSL D100, SMSL D10, Fosi Audio DS2 all using Cirrus Logic CS43131 chips...

[torgeirs]

There’s an “and” in the sentence.
So both flat frequency response and added harmonics

 

[antcollinet]

There’s an “and” in the sentence.
So both flat frequency response and added harmonics

OK - what do you mean by added harmonics.

How do flat frequency response and added harmonics result in stronger harmonics.

 

[Matt_Holland]

OK - what do you mean by added harmonics.

How do flat frequency response and added harmonics result in stronger harmonics.

He might be referring to added harmonic distortion, not that this validates his point.

 

[Philbo King]

which is all well and good. But frequency response has nothing to do with harmoncs. And all the nonsense about "stripping harmonics" is just that : Nonsense.

Except for square waves above 7 KHz. The bandpass of the reconstruction filter will remove 3rd (and higher) harmonics simply because those harmonics are not in the audio range. Google 'Gibbs Effect' if interested.

 

[Matt_Holland]

my general entry level of understanding… along with my exceptional listening skills no doubt due to being Earwax FREE for many years now.

Whether valid (lack of controls etc), you clearly have a very discerning ear and experience subtle differences between different audio equipment. I say experience because unless compared blind, you can’t prove the differences are real. In sighted comparisons I am able to “experience” differences between kit, even DACs. However, I know the differences between DACs would go away if compared blind.

My theory is that you do experience differences but you are describing them incorrectly as a change to the harmonic content or structure of the signal. As others have already mentioned, no DAC can change the harmonic structure of the original signal unless it has a non-flat frequency response. Therefore what you experience (if true) must be due to some other mechanism.

 

[torgeirs]

not that this validates his point.

Or laplaces point 200-300 years ago

 

[torgeirs]

Kind of fundemental, but takes many hours or years of maths to understand

 

[voodooless]

takes many hours or years of maths

That’s quite a vast distance between those time frames

Really though. It’s not that hard to conceptually understand why removing harmonics is nonsense:

Any music signal already includes the harmonics of the instruments. They are baked in to the waveform. There is now no way to distinguish between what is a fundamental and what is a harmonic. So removing something that you cannot even distinguish, is simply impossible.

 

[Matt_Holland]

That’s quite a vast distance between those time frames

Really though. It’s not that hard to conceptually understand why removing harmonics is nonsense:

Any music signal already includes the harmonics of the instruments. They are baked in to the waveform. There is now no way to distinguish between what is a fundamental and what is a harmonic. So removing something that you cannot even distinguish, is simply impossible.

I agree with the point you’re making, but I suppose if we’re being thorough we should state that we can remove harmonics using a low-pass filter. Of course it’s not “stripping” them out it’s just not reproducing any waveform that represents frequencies (whether fundamental or harmonic) above the low-pass cut-off.

 

[antcollinet]

Except for square waves above 7 KHz. The bandpass of the reconstruction filter will remove 3rd (and higher) harmonics simply because those harmonics are not in the audio range. Google 'Gibbs Effect' if interested.

Except that has nothing to do with the original post - plus those harmonics won't exist in the original band limited recording. They are not there to be stripped.

 

[torgeirs]

That’s quite a vast distance between those time frames

Guess it depends on what level you start reading math and how close you are to Laplace transformations

 

[torgeirs]

low-pass filter

Alters the frequency response so it is no longer flat.
So then the whole math collapses and nothing can be predicted if it is not a fixed transfer function low pass filter

 

[torgeirs]

But actually the removing of frequency thing decribes a comb filter. The same as every room will give you

 

[Matt_Holland]

Alters the frequency response so it is no longer flat.
So then the whole math collapses and nothing can be predicted if it is not a fixed transfer function low pass filter

Yes I know it alters the frequency response. My point is that a LPF can remove harmonics if they are above the cut-off frequency. Obviously not only harmonics are removed.

 

[RayDunzl]

Removing harmonics per se is a complicated process that needs to be done in the frequency domain, and even today in 2025 I'm not sure if there are any tools that will do this in real time.

I "removed" harmonics generated by a speaker playing a pure sine test tone:

Third Harmonic (not in the signal) created by JBL LSR 308

Add "anti-harmonic" to the signal

Using the Distortion controls, play with the numbers, then add third harmonic to the signal at -40 degree phase angle (depends on the phase of the harmonic).

The third from the speaker disappears.

I thought that was amusing.

Figuring out how to do it for music would be an interesting exercise.

 

[voodooless]

I thought that was amusing.

Figuring out how to do it for music would be an interesting exercise.

Well, this is more than amusing! For this to work with things like speakers, you would need to know the full dynamic behavior of the speaker to fully model the distortion, so you can then create an anti-distortion signal that works for any input. Very hard to do.

I think there are (relatively) easier ways to reduce speaker distortion, like with motion feedback (either using a sense coil, accelerometer or other means).

 

[torgeirs]

AI migth be there to help us? Who knows

 

[kemmler3D]

I "removed" harmonics generated by a speaker playing a pure sine test tone:

Third Harmonic (not in the signal) created by JBL LSR 308

View attachment 466217


Add "anti-harmonic" to the signal

Using the Distortion controls, play with the numbers, then add third harmonic to the signal at -40 degree phase angle (depends on the phase of the harmonic).

View attachment 466218

The third from the speaker disappears.

I thought that was amusing.

Figuring out how to do it for music would be an interesting exercise.

Removing fixed distortion harmonics from a speakers output is one thing... Removing harmonics from musical content is something else entirely if you're not just talking about a lowpass. Like which harmonics are we even talking about removing, and from what part of the spectrum?

As @torgiers mentions you can do stuff like this with ML / AI tools but just (say) removing the third harmonic from musical notes is not trivial like this, and I'm not sure there's a tool to do it at all, and I doubt it can be done in real-time. You need the phase, frequency, and amplitude of each harmonic you want to remove and you have to get all of that out of the raw waveform somehow, even though it's changing constantly.

 

[RayDunzl]

As @torgiers mentions you can do stuff like this with ML / AI tools but just (say) removing the third harmonic from musical notes is not trivial like this.

Yes, I think it is pretty clear that removing speaker harmonics by modifying a musical signal going to the speaker would likely be quite difficult.

Analyze the speaker at all levels, analyze the music for fundamentals that might excite the speaker's harmonics, and adjust the signal to compensate. Difficult. Yes.

However, I did demonstrate a "proof of concept" by sending a single Sine, observing the high level third harmonic the JBL LSR 308 generates on its own, then modifying that sine with an "opposing" third harmonic and having the speaker's third harmonic be Poof! Gone!

(I also consider us to be Proof of Concept for intelligent aliens, but that's another thread)

 

[MaxwellsEq]

I "removed" harmonics generated by a speaker playing a pure sine test tone:

Third Harmonic (not in the signal) created by JBL LSR 308

View attachment 466217


Add "anti-harmonic" to the signal

Using the Distortion controls, play with the numbers, then add third harmonic to the signal at -40 degree phase angle (depends on the phase of the harmonic).

View attachment 466218

The third from the speaker disappears.

I thought that was amusing.

Figuring out how to do it for music would be an interesting exercise.

Excellent experiment. I've wondered about this for decades, at which point will we have feedback loops (other than motion sensors) in loudspeakers.

BTW did it sound any different?