High-resolution Speaker Frequency Response & Distortion Measurements and Why We Need Them (video)

[dc655321]

"..A speaker system or driver is a high-pass filter. That is why they cannot reproduce DC..."

Yes but i think its much more easy to understand, that a speaker that not moves cant produce sound. In case of DC on a speaker.

Yes.
SPL is proportional to acceleration.
No acceleration, no SPL.

 

[Owl]

Thanks for the detailed explanation. I love the educational videos. I see speaker manufactures put out test for their drivers, how would one predict enclosure responses after installation? The cabinet will surely change everything.

 

[tomtoo]

Thanks for the detailed explanation. I love the educational videos. I see speaker manufactures put out test for their drivers, how would one predict enclosure responses after installation? The cabinet will surely change everything.

Math is your friend. For this simulations exist. At least they can give a idear.

 

[lashto]

The electronics guys have been making and selling that gear for years without proper measurements as well. Yet I keep hearing left and right that they are now buying $30,000 Audio Precision analyzers to make sure they can make proper measurements. Lest they want us to be the first who does and sees problems.

Just as well, I expect speaker manufacturers, large and small, to buy Klippel NFS. If they already have the baseline Klippel hardware and don't don't need all the bells and whistles, an NFS add-on can be $50K or less. We have seen the beginning of this trend and I am confident it will grow rapidly with time. The value proposition is unique and quite high.

The old rumor has it that Amir/ASR are Topping salesmen.
You just started a new rumor: Amir is a Klippel/AP salesman

That is progress

 

[Robbo99999]

@amirm , thanks for the vid, I think I understood most of it, but I didn't understand the following bit:

I really didn't understand that graph in how the frequency response was above time zero, and how the distortion harmonics were at negative time - as I thought distortion was initiated at "zero time" when the frequency was played back on the speaker, and then the speaker would vibrate and have resonances that would "echo" after that point (ie "plus time"), creating audible artifacts both during (zero time) & maybe also shortly after the initial event (plus time). I also thought that the distortion harmonics were due to where on the frequency response the distortion is audible or measurable - so I thought 3rd Harmonics for example meant that if your speaker had an issue say at 30Hz, as in getting stressed playing 30Hz then that can create audible artifacts at 30Hz x 3, so it would create an audible artifact at 90Hz as a result of the initial 30Hz playback tone? I guess I don't understand that graph and I may not understand harmonic distortion properly either. I also don't quite understand how the microphone detects different harmonic distortions....going back to my 30Hz distortion example, if it's 3rd Harmonic again, so does that mean the mic is measuring what it hears when the 30Hz tone is played during the chirp, and then at the same "exact" time it detects the 90Hz artifact that would be the 3rd Harmonic Distortion created from the 30Hz tone.....and also would that be described as 30Hz 3rd Harmonic Distortion or would that be shown in the graphs as 3rd Harmonic Distortion at 90Hz? I definitely don't have a proper handle on this part of your presentation and the other questions I asked, would you be able to put it into a framework of understanding for me (us) - maybe you have some links that answers it, or you know how to explain what I'm after with a few paragraphs or some more graphs?

 

[NTK]

@amirm , thanks for the vid, I think I understood most of it, but I didn't understand the following bit:
View attachment 216427
I really didn't understand that graph in how the frequency response was above time zero, and how the distortion harmonics were at negative time - as I thought distortion was initiated at "zero time" when the frequency was played back on the speaker, and then the speaker would vibrate and have resonances that would "echo" after that point (ie "plus time"), creating audible artifacts both during (zero time) & maybe also shortly after the initial event (plus time). I also thought that the distortion harmonics were due to where on the frequency response the distortion is audible or measurable - so I thought 3rd Harmonics for example meant that if your speaker had an issue say at 30Hz, as in getting stressed playing 30Hz then that can create audible artifacts at 30Hz x 3, so it would create an audible artifact at 90Hz as a result of the initial 30Hz playback tone? I guess I don't understand that graph and I may not understand harmonic distortion properly either. I also don't quite understand how the microphone detects different harmonic distortions....going back to my 30Hz distortion example, if it's 3rd Harmonic again, so does that mean the mic is measuring what it hears when the 30Hz tone is played during the chirp, and then at the same "exact" time it detects the 90Hz artifact that would be the 3rd Harmonic Distortion created from the 30Hz tone.....and also would that be described as 30Hz 3rd Harmonic Distortion or would that be shown in the graphs as 3rd Harmonic Distortion at 90Hz? I definitely don't have a proper handle on this part of your presentation and the other questions I asked, would you be able to put it into a framework of understanding for me (us) - maybe you have some links that answers it, or you know how to explain what I'm after with a few paragraphs or some more graphs?

The method used to separate out the individual harmonic distortions is described in this paper by Professor Farina.

https://www.researchgate.net/publication/277293870_Simultaneous_measurement_of_impulse_response_and_distortion_with_a_swept-sine_technique_Audio_Engineering_Society_Convention_108_Audio_Engineering_Society

 

[PeteL]

"..A speaker system or driver is a high-pass filter. That is why they cannot reproduce DC..."

Yes but i think its much more easy to understand, that a speaker that not moves cant produce sound. In case of DC on a speaker.

I am not sure why the need for that fundamental theoretical stuff. OK maybe a pure square wave is not an audio signal, driver limitations or not, it requires an infinite numbers of harmonics to have a square wave and not only the driver can't do that whether we want to call it DC or not but the ears can't hear that. But It's just deviating from the subject of audio reproduction, we already know that it is theoretically impossible to reproduce a pure Square, but the point is in audio, a band limited square wave, is simply called a square wave, Audio Synthetisers have been working with Square waves for ever, we all know what a square wave sounds like, bottom line some transducer will play them with more or less fidelity, but it's what it is, I want a 1kHz squarewave to sound like a Square wave, even imperfect, not a Sawtooth. It's an audio signal it just have limitations and will never be pure DC when applied to audio.

 

[Axo1989]

Thanks for clarifying the issue with square waves & acoustical driver measurements. I have no audio-related qualifications but that sounds simple enough (even) for me.

However, when you say "a rectangle its the bad behaviour of the driver" it implies that a measurable "good behavior of the driver" exists too. Identifying the good/bad components is the whole point of those reviews/measurements, so it seems that square waves might actually be an useful test.
Or am I missing something?!

You might enjoy this article, which describes correcting a speaker response wrt square wave by applying first frequency then phase correction. I often find working through a practical/applied scenario easier to follow.

 

[tomtoo]

I am not sure why the need for that fundamental theoretical stuff. OK maybe a pure square wave is not an audio signal, driver limitations or not, it requires an infinite numbers of harmonics to have a square wave and not only the driver can't do that whether we want to call it DC or not but the ears can't hear that. But It's just deviating from the subject of audio reproduction, we already know that it is theoretically impossible to reproduce a pure Square, but the point is in audio, a band limited square wave, is simply called a square wave, Audio Synthetisers have been working with Square waves for ever, we all know what a square wave sounds like, bottom line some transducer will play them with more or less fidelity, but it's what it is, I want a 1kHz squarewave to sound like a Square wave, even imperfect, not a Sawtooth. It's an audio signal it just have limitations and will never be pure DC when applied to audio.

Hard to say how a perfect square wave should sound when a speaker cant do them? More harmonics? Could be just failures from the speakers? Different FR different sound of the square. I not see how a square should be a good signal to judge a speaker?

 

[gnarly]

Hard to say how a perfect square wave should sound when a speaker cant do them? More harmonics? Could be just failures from the speakers? Different FR different sound of the square. I not see how a square should be a good signal to judge a speaker?

If by perfect square wave, you mean a theoretical square wave that includes an infinite series of harmonics, of course a speaker can't reproduce that.
But neither can electronics...they too have a finite bandwidth .....the infinite part makes "can't reproduce" a theoretical 'so what' imo.

Speakers can replicate excellent square waves within their audio range bandwidth limitations.
If you will look at the set I posted in #55, you can see the lower frequency square waves rise quite fast, as they have more available harmonics within the audio spectrum.
In contrast to 1000Hz where they don't have enough harmonics to let them rise steeply. But again, so what. 1000Hz and up can still be reproduced as good as possible given audio's bandwidth.

The simple clincher in my mind is this: perfect frequency response (mag and phase) = a perfect set of square waves across the spectrum, until HF harmonics aren't available.

They are identities, just like perfect mag and phase = perfect impulse = perfect step response = perfect squares waves within their pragmatic frequency range
Square waves are still an excellent way to measure/judge a speaker...it's just FFT's mag and phase are a whole lot easier to understand, and act on.

 

[amirm]

I really didn't understand that graph in how the frequency response was above time zero, and how the distortion harmonics were at negative time - as I thought distortion was initiated at "zero time" when the frequency was played back on the speaker, and then the speaker would vibrate and have resonances that would "echo" after that point (ie "plus time"), creating audible artifacts both during (zero time) & maybe also shortly after the initial event (plus time).

This is a mathematical concept. Obviously there is no such thing as negative time. As an analogy, a phase shift in electronics is represented by imaginary numbers. Phase shift is of course real but we represent it using something imaginary for the sake of math.

 

[amirm]

That is something I did not expect to hear from you. There is quite a bunch out there who keeps saying that "test sinewaves are not audio/music", do you want to spend your week talking to them ?!
And those 'chirps' aren't exactly audio/music either ...

That was not the point of my comment. For anything to resemble square wave, you need to have tons and tons of bandwidth. If you say, ship that as 44.1 kHz CD, you are just not going to get that due to band limiting of CD. A chirp signal does not need bandwidth beyond its maximum frequency so definitely is better in this regard. Indeed you can listen to a Chirp and detect problems in the device by ear -- something you can't do with square wave.

 

[PeteL]

Hard to say how a perfect square wave should sound when a speaker cant do them? More harmonics? Could be just failures from the speakers? Different FR different sound of the square. I not see how a square should be a good signal to judge a speaker?

I don't have the expertise in speaker measurments, but do we agree that a Square wave source, reproduced by a driver, picked up by a measurment microphone should look closer to a square wave than a Sawtooth? So to answer that, Yes, I assume a faster driver, would allow to maintain more harmonics and not reset to position like with a impulse since we are fundamentally in the time domain with drivers, how relevent of a measure of performance, My knowledge is too limited, but what I do know is I know what a square wave sounds like, it's very different than a sawtooth, more nasal and hollow. so it shouldn't measure the same. Of course it only applies for lower frequency fundamentals. A 7K square wave will sound like a sine, but not a 500 Hz one,

 

[tomtoo]

I don't have the expertise in speaker measurments, but do we agree that a Square wave source, reproduced by a driver, picked up by a measurment microphone should look closer to a square wave than a Sawtooth? So to answer that, Yes, I assume a faster driver, would allow to maintain more harmonics and not reset to position like with a impulse since we are fundamentally in the time domain with drivers, how relevent of a measure of performance, My knowledge is too limited, but what I do know is I know what a square wave sounds like, it's very different than a sawtooth, more nasal and hollow. so it shouldn't measure the same. Of course it only applies for lower frequency fundamentals. A 7K sine wave will sound like a sine, but not a 500 Hz one,

".....but do we agree that a Square wave source, reproduced by a driver, picked up by a measurment microphone should look closer to a square wave than a Sawtooth?..."

No i dissaggree.

Imagine, very easy measurementapparatus:

Sender:Batterie 5v, switch, speaker

Microphone(magic mic, gives same voltage as speaker in), Voltmeter(Oszi)

What happens when you switch on?
Speaker goes to 5v position. And stays there. Microphone goes to -5V position and goes back to 0V position. You see on the receiver side your nice half of the rectangle is something like a reversed sawtooth. The speaker can do a rectangel, but the mic cant register that. Couse the top of the rectangel is 0V at the mic. But i repeat myself, you can see this in the steprespons.

 

[lashto]

That was not the point of my comment. For anything to resemble square wave, you need to have tons and tons of bandwidth. If you say, ship that as 44.1 kHz CD, you are just not going to get that due to band limiting of CD. A chirp signal does not need bandwidth beyond its maximum frequency so definitely is better in this regard. Indeed you can listen to a Chirp and detect problems in the device by ear -- something you can't do with square wave.

Still dont think that we have any disagreement there
Actually, I don't evan think that a test signal should be audio/audible. It just needs to identify a technical issue/aspect. Like those KHz/MHz spikes that Class-D and SigmaDelta chips produce. Don't care if those are audible or not, but would like to see them.

I just wouldn't use the "it's not audio" argument/wording .. unless you want more 'smoke' from the "it's not audio" crowd.

 

[PeteL]

".....but do we agree that a Square wave source, reproduced by a driver, picked up by a measurment microphone should look closer to a square wave than a Sawtooth?..."

No i dissaggree.

Imagine, very easy measurementapparatus:

Sender:Batterie 5v, switch, speaker

Microphone(magic mic, gives same voltage as speaker in), Voltmeter(Oszi)

What happens when you switch on?
Speaker goes to 5v position. And stays there. Microphone goes to -5V position and goes back to 0V position. You see on the receiver side your nice half of the rectangle is something like a reversed sawtooth. The speaker can do a rectangel, but the mic cant register that. Couse the top of the rectangel is 0V at the mic.

Well, closing the loop again, I am not talking about intermitent DC, again, to be clear, a 250 Hz Square wave sounds very different than a 250 Hz Sawtooth. 500 Hz too. It's obvious, if it's not the reproduction should be questioned. It's NOT A rectangle, It's not DC, It only is in fundamental maths. It's a fundamental with odd harmonic content, band limited.

 

[Axo1989]

".....but do we agree that a Square wave source, reproduced by a driver, picked up by a measurment microphone should look closer to a square wave than a Sawtooth?..."

No i dissaggree.

Imagine, very easy measurementapparatus:

Sender:Batterie 5v, switch, speaker

Microphone(magic mic, gives same voltage as speaker in), Voltmeter(Oszi)

What happens when you switch on?
Speaker goes to 5v position. And stays there. Microphone goes to -5V position and goes back to 0V position. You see on the receiver side your nice half of the rectangle is something like a reversed sawtooth. The speaker can do a rectangel, but the mic cant register that. Couse the top of the rectangel is 0V at the mic. But i repeat myself, you can see this in the steprespons.

Well, if you wait half an hour before you switch off, your "square wave" is very low frequency. Below my audible range anyway, maybe if I was a blue whale?

 

[tomtoo]

Well, closing the loop again, I am not talking about intermitent DC, again, to be clear, a 250 Hz Square wave sounds very different than a 250 Hz Sawtooth. 500 Hz too. It's obvious, if it's not the reproduction should be questioned. It's NOT A rectangle, It's not DC, It only is in fundamental maths. It's a fundamental with odd harmonic content, band limited.

Yes it sounds different. But it sounds different is not a measurement. And sure is a rectangel DC that a speaker cant produce acusticly. You measure not a rectangel on that microphone. And thats absolutly right.
Tell me how you should measure a rectangel on the microphone?

 

[tomtoo]

Well, if you wait half an hour before you switch off, your "square wave" is very low frequency. Below my audible range anyway, maybe if I was a blue whale?

A DC has no frequence.

 

[PeteL]

A DC has no frequence.

What do you mean by that? The whole computing world is based on this very concept. a fixed voltage alterning states at a known frequency.