Speakers distortion

[Krunok]

The distortions (harmonics) rise faster than the fundamental as the speaker exceeds its comfort zone and the fundamental level increases.

I'm not sure I understand what you mean with "rise faster"?

They may or may not be recognized with music - I think it disturbs the "clarity" of the presentation. I don't perceive my little JBLs to be as "clean" as the electrostats at moderately high levels with music.

I agree - "clarity" is the word I would use as well. "Fidelity" also springs to mind..

Harmonic distortion does not visually affect the frequency response measurement (much), because, the harmonic level is (typically) lower than the measured fundamental across the frequency band. Even if the harmonic "adds" to the level reported the sum is small.

Assume 80dB fundamental with a -10dB second harmonic (severe, easily audible on a tone), with additional tones at -20, -30, -40, etc:

Just calculating SPL here, an 80dB main tone, and these severe harmonics contribute only another .45dB:


View attachment 19840

JBL vs ML at 82dB at the listening position.

View attachment 19841

I believe it doesn't matter that total level is changed but harmonics change a shape of the wave and ear somoehow catches it.

Qutie a difference though between JBL and MLs. Do you get consistent figures when you repeat the measurement?

 

[Blumlein 88]

I'm not sure I understand what you mean with "rise faster"?


I agree - "clarity" is the word I would use as well. "Fidelity" also springs to mind..

I believe it doesn't matter that total level is changed but harmonics change a shape of the wave and ear somoehow catches it.

Qutie a difference though between JBL and MLs. Do you get consistent figures when you repeat the measurement?

At 100 db spl distortion might be 40 db lower.
At 90 db spl distortion might be 50 db lower.
At 80 db spl distortion might be 70 db lower.

So you can't eq out distortion as the level of it rises faster than spl rises. Meaning the effect on measured FR varies with output level.

 

[RayDunzl]

I'm not sure I understand what you mean with "rise faster"?

A speaker (assume perfect amplification) oscillates. Imperfections in the speaker (not an ideal piston or whatever) induce flapping/bending motions, generally harmonically related to the fundamental, with different levels of "flap" at various frequencies (depending on whatever).

Push the speaker harder, the "flapping" gets worse. Reduce the level of the fundamental, the flapping may even appear to disappear.

JBL 305P MKII
Three fundamental levels 70dB, 80dB, and 90dB, two speakers playing, at the listening position (10 feet)

Second harmonic rises faster than the fundamental, going from about 19dB to 41dB - difference 22dB

Second harmonic rises from about 42dB to 63dB - difference 21dB

So, originally speaking loosely, the "harmonics rise faster than the fundamental" could be taken as true (2nd harmonic in this case), or false (3rd harmonic didn't cooperate with my hypothesis).

It's something I've noticed but not specifically measured before.

My use of distortion measurements is for personal information (and amateur posts like this), pass/fail, and not trying to determine exact levels for any useful purpose.

 

[Krunok]

At 100 db spl distortion might be 40 db lower.
At 90 db spl distortion might be 50 db lower.
At 80 db spl distortion might be 70 db lower.

So you can't eq out distortion as the level of it rises faster than spl rises. Meaning the effect on measured FR varies with output level.

We have already said that correction should be made on "distortion response" table which itakes into account not only frequency but also a level of the fundamental tone.

 

[RayDunzl]

Qutie a difference though between JBL and MLs. Do you get consistent figures when you repeat the measurement?

Let's see. Martin Logan panels:

70dB

80dB

90dB

 

[Krunok]

So, originally speaking loosely, the "harmonics rise faster than the fundamental" could be taken as true (2nd harmonic in this case), or false (3rd harmonic didn't cooperate with my hypothesis).

Sure, THD figures certainly depends on the level, but as I said, the idea was from the beginning to take that into account.

 

[RayDunzl]

Qutie a difference though between JBL and MLs. Do you get consistent figures when you repeat the measurement?

If you meant "repeating the measure on the JBL", yes, consistent (in-room repeats are never exact, and may vary a little as you look at them).

 

[DonH56]

We have already said that correction should be made on "distortion response" table which itakes into account not only frequency but also a level of the fundamental tone.

That is not the same thing as EQ'ing out the distortion; that only corrects for the room response. It will be valid only for one measurement condition (mic position, volume, frequency, etc.) and will not work at any other condition nor in general in the presence of other signals.

 

[Blumlein 88]

Had some time to try out my idea about getting genuine speaker distortion levels in room. My goal was to find out if I could determine speaker distortion as I would get were this done outdoors or anechoically. I did a rough and ready test. As often is the case I answered some questions and found a new one to investigate. For clarity and quickness at the moment I only looked at 2nd and 3rd harmonic distortion.

Using an LSR305 with no particular positioning, simply doing RTA averages of a single tone at 440 hz I did three positions getting a raw measure of -54 db, -41 db and -52 db 2nd harmonic relative to the main tone. No compensation. For 3rd it was -49 db, -51 db and -52 db. Using a sweep measure gave me 52,50, and 54 for 2nd and 50, 50, and 56 for 3rd. Sorry should have made a table of this.

Next I started by adding a 2nd harmonic to the test tone at -40 db and comparing vs no added tone. I adjusted it and found -48 db caused the 2nd harmonic to go down by 6 db when the extra tone was taken away. My thinking being this is where my added 2nd harmonic equaled real distortion. As I had hoped the same amount was correct for all three measuring positions. The variance doing this a few times at each position was only about plus or minus a db or so. Repeated the same procedure with 3rd harmonic. It worked with a 3rd harmonic added at -47 db. You did have to invert the added tone or it cancelled out the speaker distortion causing the 3rd harmonic to rise when the extra tone stopped.

Okay so far so good. So next I tried the moving microphone method. I first did it in a sloppy manner with a sweep measurement. I rotated the microphone in my hand in an oval that was about 2 feet long and 1ft tall and canted from perpendicular to the speaker at various angles. I rotated a bit more than 1 rotation per second so it was 6 or 7 rotations in the 5.5 second sweep I was using. I thought this was hopeless. But for 2nd harmonic the results at the 440 frequency were 48.4 db, 49 db, and 46.8 db. Hmmmm, not so bad. For 3rd harmonic it was 47 db, 47.2 db and 48.1 db. Also not so bad. I did this a few more times since it was so easy, and results stayed right in these narrow ranges.

So next I used the moving microphone method with a single tone with the REW RTA. It was very consistent, but here we have the new puzzle. Letting it do a 32 FFT average and letting it run plenty long I could watch the distortion products to see if they were settling down. They were. At just about exactly 6 db lower than the sweep measures with my goofy MMM. Converting the percentages it was bouncing around -54 db for second harmonic and about -53 or 52 db for 3rd harmonic.

So moving works I think. Why does a single tone test 6 db better?

Now this was quick and dirty as I was short on time, but bears further investigating. At more frequencies too. So which of the MMM results are correct, and why the 6 db difference? Will it hold up at other frequencies? Maybe when I get a chance to test outdoors I can see if these are close to the intrinsic distortion values for the speaker.

Of course if speakers generally have distortion at .5% or less on tones it probably is of no consequence with music.

 

[RayDunzl]

Why does a single tone test 6 db better?

Could you summarize what leads to that question?

My reading comprehension neuron may not be working well tonight.

(I'm exercising it some more right now. Maybe getting closer)

 

[Krunok]

That is not the same thing as EQ'ing out the distortion; that only corrects for the room response. It will be valid only for one measurement condition (mic position, volume, frequency, etc.) and will not work at any other condition nor in general in the presence of other signals.

Why not? I think it would correct for a distortion harmonics affected by room.

Are you familiar how convolution engine corrects frequency response based on your measurement? This is the same thing except that level of the fundamental tone would be taken into account and 2nd and 3rd distortion harmonic would be canceled.

 

[Krunok]

I rotated the microphone in my hand in an oval that was about 2 feet long and 1ft tall and canted from perpendicular to the speaker at various angles. I rotated a bit more than 1 rotation per second so it was 6 or 7 rotations in the 5.5 second sweep I was using.

Very interesting experiment!

Btw, I'm not sure I understand how you did moving microphone method. You shouldn't rotate microphone in your hand, you should move your hand with the microphone steady in it. Try to keep steady speed of app 10cm/sec. You should also perform movements in all 3 geometrical plains or you can try to make movements like you are tryign to connect those green dots in Dirac Live scheme with your mic.
If you want to record higher frequencies with your method you should also keep microphone aimed at the speaker all the time.

When I was doing frequency response measurements the response I got from RTA was also lower than the one I got from sweeps for similar ammount. It is how REW does it, I have no idea why. Maybe it's even a bug.

 

[Blumlein 88]

Could you summarize what leads to that question?

My reading comprehension neuron may not be working well tonight.

(I'm exercising it some more right now. Maybe getting closer)

Using a moving microphone method with a single tone averaged 32 times on the REW RTA resulted in consistent readings at all positions.

Using a moving microphone method with sweeps resulted in consistent readings at all positions.

Using a fixed microphone at these positions and adding harmonics to ferret out distortion level lead to consistent results at all positions on the RTA.

So using fixed position single tones with RTA and MMM sweeps gave me the same numbers and consistent at 3 positions.

But using single tone RTA with the MMM gave consistent results 6 db lower than the above two methods.

 

[Blumlein 88]

Very interesting experiment!

Btw, I'm not sure I understand how you did moving microphone method. You shouldn't rotate microphone in your hand, you should move your hand with the microphone steady in it. Try to keep steady speed of app 10cm/sec. You should also perform movements in all 3 geometrical plains or you can try to make movements like you are tryign to connect those green dots in Dirac Live scheme with your mic.
If you want to record higher frequencies with your method you should also keep microphone aimed at the speaker all the time.

My method was like me rather sloppy. I did attempt to maintain aim at the speaker. But I was literally whirling it about in my hand at about one oval rotation per second. Faster than 10 cm/second. More like 10 times that fast.

On the single tones you can watch the distortion readout in REW. It settled right down and varied less than .05% over several seconds.

 

[RayDunzl]

Jeez... I must be in bad shape tonight... re-reading the summary now...

---

Dammit! I thought I was good, then the last sentence threw me. Starting over...

 

[Krunok]

My method was like me rather sloppy. I did attempt to maintain aim at the speaker. But I was literally whirling it about in my hand at about one oval rotation per second. Faster than 10 cm/second. More like 10 times that fast.

Still an interesting experiment though!

Here's a simple way to make MMM:

- start at the center point of your LP and move mic like you are drawing symbol for eternity (number 8 rotated by 90 degrees) in the vertical left-right plane. Width of the symbol should be 1m and height 40-50cm.

- then draw a circle in the horizontal forward-backward plane of app 70cm diameter (center of the circle being cneter point of your LP).

- then draw a circle in the vertical up-down plane of app 50cm diameter (center of the circle being cneter point of your LP)

Do not move mic faster than 10cm/sec, keep it steady in your hand pointed at the speakers. That should do it!

 

[Cosmik]

A very interesting thread that focuses the mind. What it tells me:

If you test at a fixed frequency, and you assume the room is linear, you know that any harmonics you measure are distortion. Of course the room may not be linear: the harmonics may come from vibrating objects or even flexing panels within the room.

Room reflections will mix with the fundamental and harmonics from the speaker and affect their measured amplitude, sometimes boosting the amplitude and sometimes reducing it. Any single measurement that includes reflections cannot tell you the proportion of distortion products relative to fundamental.

Multiple tests averaged with different mic positions and/or slightly different fundamental frequencies will probably improve the accuracy, but only by some unknown amount.

Near field measurement would be good because it reduces room effects but would only work properly at low test tone frequencies and low harmonic frequencies (wavelengths long compared to driver diameter), otherwise the fundamental and/or distortion products are attenuated (but possibly by a predictable and therefore compensate-able amount..?). To work best it would need a specialised microphone designed for high SPLs, otherwise it just trades off room effect uncertainty for higher microphone distortion.

There are several speaker distortion mechanisms, including cone break-up and Doppler distortion.

Distortion products caused by cone break-up will have nonlinear 'onset' and 'hysteresis' and their own peculiar dispersion characteristics. Shifts in test frequency and/or amplitude may cause these characteristics to change in unknown ways. These characteristics may be temperature and humidity dependent.

Conclusions:
Accurate measurements can only be made in properly anechoic conditions, but as to whether they are meaningful given the nature of cone break-up type distortion...

Never mind the difficulties in making meaningful measurements. What this says to me is that a notion I have often had about pre-distorting the signal using a neural network to compensate for speaker distortion is pie in the sky and wouldn't work, not least because I couldn't meaningfully create the training data and measure the results.

By far the most significant conclusion is that the speaker should be designed to avoid distortion in the first place. Designing a low distortion speaker would be much easier than measuring a speaker for distortion!

There are obvious methods for reducing speaker distortion. These methods may clash with traditional speaker design rules of thumb e.g. "Never place a crossover in the middle of the vocal range". Thankfully active DSP allows you to do this far more transparently than was ever possible previously. You probably also need to make space in your living room for a box that's a bit bigger than is fashionable.

 

[Krunok]

Any single measurement that includes reflections cannot tell you the proportion of distortion products relative to fundamental.

Multiple tests averaged with different mic positions and/or slightly different fundamental frequencies will probably improve the accuracy, but only by some unknown amount.

With RTA MMM it's the same as with frequency response - if you can get consistent (repeatable) results than your accuracy should be fine.

Let me offer my view of the in-room measurement: you play a 30Hz sine tone at say 60dB. You measure SPL at that tone to see how different it is from 70dB (frequeny response). Then you measure 2nd and 3rd harmonics outcome of that tone (distortion response). You repeat that measurement for the same frequency at 65dB, 70dB, 75dB, .. 105dB. You do the same for other frequencies.

Yep, you get quite a large table, but I see no reason why it couldn't be used to correct for distortion harmonics (2nd and 3rd). Quesiton is of course, how you gonna do it? With frequency response you use FIR but what is the math to cancel harmonics? If you solve that you got yourself a simple distortion cancellation system that will work in a similar way as frequncy response EQ does (without realtime feedback but with premeasured set of correction parameters).

 

[svart-hvitt]

A very interesting thread that focuses the mind. What it tells me:

If you test at a fixed frequency, and you assume the room is linear, you know that any harmonics you measure are distortion. Of course the room may not be linear: the harmonics may come from vibrating objects or even flexing panels within the room.

Room reflections will mix with the fundamental and harmonics from the speaker and affect their measured amplitude, sometimes boosting the amplitude and sometimes reducing it. Any single measurement that includes reflections cannot tell you the proportion of distortion products relative to fundamental.

Multiple tests averaged with different mic positions and/or slightly different fundamental frequencies will probably improve the accuracy, but only by some unknown amount.

Near field measurement would be good because it reduces room effects but would only work properly at low test tone frequencies and low harmonic frequencies (wavelengths long compared to driver diameter), otherwise the fundamental and/or distortion products are attenuated (but possibly by a predictable and therefore compensate-able amount..?). To work best it would need a specialised microphone designed for high SPLs, otherwise it just trades off room effect uncertainty for higher microphone distortion.

There are several speaker distortion mechanisms, including cone break-up and Doppler distortion.

Distortion products caused by cone break-up will have nonlinear 'onset' and 'hysteresis' and their own peculiar dispersion characteristics. Shifts in test frequency and/or amplitude may cause these characteristics to change in unknown ways. These characteristics may be temperature and humidity dependent.

Conclusions:
Accurate measurements can only be made in properly anechoic conditions, but as to whether they are meaningful given the nature of cone break-up type distortion...

Never mind the difficulties in making meaningful measurements. What this says to me is that a notion I have often had about pre-distorting the signal using a neural network to compensate for speaker distortion is pie in the sky and wouldn't work, not least because I couldn't meaningfully create the training data and measure the results.

By far the most significant conclusion is that the speaker should be designed to avoid distortion in the first place. Designing a low distortion speaker would be much easier than measuring a speaker for distortion!

There are obvious methods for reducing speaker distortion. These methods may clash with traditional speaker design rules of thumb e.g. "Never place a crossover in the middle of the vocal range". Thankfully active DSP allows you to do this far more transparently than was ever possible previously. You probably also need to make space in your living room for a box that's a bit bigger than is fashionable.

I believe your conclusion resembles @Floyd Toole ’s previous comments?

Having said that, anecdotally, bass crazy audiophiles (of the systematic type) talk a lot about distortion in lower frequencies and invest tremendous time and efforts to lower distortion from 10 Hz and up to mid bass. Is speaker distortion mainly a lower frequency problem?

 

[Cosmik]

Let me offer my view of the in-room measurement: you play a 30Hz sine tone at say 60dB. You measure SPL at that tone to see how different it is from 70dB (frequeny response). Then you measure 2nd and 3rd harmonics outcome of that tone (distortion response). You repeat that measurement for the same frequency at 65dB, 70dB, 75dB, .. 105dB. You do the same for other frequencies.

Yep, you get quite a large table, but I see no reason why it couldn't be used to correct for distortion harmonics (2nd and 3rd). Quesiton is of course, how you gonna do it? With frequency response you use FIR but what is the math to cancel harmonics? If you solve that you got yourself a simple distortion cancellation system

This is the steady state response view of audio, but unfortunately it isn't going to work like that. In reality, what a cone will do when breaking up is highly complex, involving the history of the complex signal. Combine this with the corresponding complex dispersion effects and I think it is pointless to even try compensating for it. A small change of temperature and you would almost certainly be making it worse, anyway!