Multitone Distortion, Measurement and Evaluation

[Finn]

I would like to start a discussion on Multitone Distortion (MD) in Loudspeakers. Harmonic Distortion (HD) is well understood these days and in most speaker designs it is not a problem any more. Other than distortion products that occur when using a multitone signal for measurements. This is good explanation by purifi on different factors that create distortion inside a transducer other than (HD). This is a review from erins audio corner, I believe he is the only one measuring MD? In the last part there are graphs showing the MD from 80 Hz to 20 kHz and 20 Hz to 20 kHz at various levels. Unfortunately it is not really known what exactly the signal looks like?
I mainly use arta to do loudspeaker measurements (it is free now!!), it makes it possilble to use different stimuli. I would be curios which would be good and how to interpret the values? Interesting would be two tone signals or multitone signals. With multitone signals there would be the question how many tones per octave and if a filter should be applied to the singnal?

Another topic would be audilbility of MD? How should a level of MD be evaluated?

 

[Finn]

I have watched Jospeh Crowes Video about MD Distortion and he sets 1% / -40 dB as a good value for MD Distortion. Further he also says that when evaluating at a certain SPL, 86 dB or 96 dB it is a RMS value. This means that If you listening to music at 96 dB the crest factor plays an important role. The transients can have 106+ dB depending on the material.

 

[Finn]

And I had forgotton Erin's video about Distortion. It rounds up quiet a lot and gives good listening comparisons of the different types of distortion.

 

[Heinrich]

And I had forgotton Erin's video about Distortion. It rounds up quiet a lot and gives good listening comparisons of the different types of distortion.

Only to not leave you with no response. I experienced the same, people are stuck with the spinorama, speculating a lot of crazy stuff. Won't tell, but there is more, simply put. I just switched back from spin-perfect speakers to a massive p/a oriented design, and whoohey, shocking. Way more lean and easy in bass, but so satisfying, bass lines in shape and rumble, clarity and distinction in the vocals, revelations in the treble way beyond a 3/4th inch tweeter while being +/- 1dB all across the board. In short a marvel for the recorded music lover, speakers to forget, definitely true to the source. And then, I know they might get 4 times as loud with no stress other than having the neighbors sitting on my lap.

Look, many use 'science' as a badge to attach to their "system", very much like others trusting the glossy magazines. There's little more - my very personal impresson - uups - which is for sure not true.

Now I have the opportunity: artist Noname, Sundial, the album. It feels like physical therapy, there's a touch to it, from scratching hair to bending muscles, pause and icy showers included. That's not in the spinorama, but is art in recorded music.

 

[dfuller]

Multitone distortion is hands down more audible than harmonic distortion. It's pretty much always better to have a 3 way than a 2 way for IMD for that reason - unless you have a really high end woofer (see: Purifi) it's always going to be problematic if you have much of any bass extension in a typical two way.

In basically every two way standmount Erin has MD data for, it's always peaking right before the crossover. This often doesn't happen with 3 ways, solely because the woofer isn't trying to recreate signals 6-7 octaves apart from one another.

 

[Heinrich]

It's pretty much always better to have a 3 way than a 2 way for IMD for that reason - unless ...

Unfortunately, I don't see any way of popularizing the topic at the moment. It will take a lot more to understand it. The capacity for this does not seem to be available. Or the motivation is lacking, reasons can be thought of.

There was already a video circulating that tried to prove the equivalence of a KEF LS50 with a JBL cinema system, one of the better ones. The author said that was the case. Well ... I think that the LS50 is quite a flop, and so with me it disappeared back into the packaging within 2 minutes. Yesterday I (also) heard a modern e-bass (5-string) with my old junk (45 years old cinema) - phew, it's awesome! The opposite of the usual expectation, not overpowering, but quietly insistent. You can't teach our ‘critical listeners’ this quality of experience. What a pitty.

 

[Finn]

Multitone distortion is hands down more audible than harmonic distortion. It's pretty much always better to have a 3 way than a 2 way for IMD for that reason - unless you have a really high end woofer (see: Purifi) it's always going to be problematic if you have much of any bass extension in a typical two way.

In basically every two way standmount Erin has MD data for, it's always peaking right before the crossover. This often doesn't happen with 3 ways, solely because the woofer isn't trying to recreate signals 6-7 octaves apart from one another.

Preferably having a 3 way instead of a 2 way is better for Multi Tone Distortion.

I honestly don't get the hype for purify woofers. If you compare the Ascilab C6B and Buchardt A10 I would pick the C6B. Since the A10 plays much lower than the C6B I picked the 80 Hz to 20 kHz from Erin's audiocorner to make better comparison. I like all the science based developments they advertise but I do not see the results in the MD Plots. Seems like in that scenario the SB Acoustics driver is better.

 

[dfuller]

Preferably having a 3 way instead of a 2 way is better for Multi Tone Distortion.

I honestly don't get the hype for purify woofers. If you compare the Ascilab C6B and Buchardt A10 I would pick the C6B. Since the A10 plays much lower than the C6B I picked the 80 Hz to 20 kHz from Erin's audiocorner to make better comparison. I like all the science based developments they advertise but I do not see the results in the MD Plots. Seems like in that scenario the SB Acoustics driver is better.

View attachment 457301
View attachment 457302

That Buchardt definitely isn't the best showing for that woofer vs this Radiance Clarity 6.2. Even so, the MD does still climb right around the crossover at higher excursions (but, true to form, it is pretty low - considering it's a 6.5" driver, staying below 3% TD at 96dB is pretty good! Many in that size class are getting closer to 10%).

As far as the Ascilab, it has a 1100hz 4th order crossover. That's why it lacks the boost in the 1-2khz range, the tweeter has already taken over.

The Audio First Fidelia has the same series of woofer as the Ascilab (IIRC) but crosses higher with a shallower filter slope. The MD is much higher.

 

[Heinrich]

That Buchardt definitely isn't the best showing for that woofer vs this Radiance Clarity 6.2. Even so, the MD does still climb right around the crossover at higher excursions (but, true to form, it is pretty low - considering it's a 6.5" driver, staying below 3% TD at 96dB is pretty good! Many in that size class are getting closer to 10%).

Don't know if Erin's data comprises Doppler. But anyway, Doppler converts to IM in-room, see

Linkwitz / Frontiers#J Doppler

That's why the Purify example

Purify Doppler tech/ note

works over headphones, but not with my virtually IM free speakers.

 

[kharan]

Preferably having a 3 way instead of a 2 way is better for Multi Tone Distortion.

I honestly don't get the hype for purify woofers. If you compare the Ascilab C6B and Buchardt A10 I would pick the C6B. Since the A10 plays much lower than the C6B I picked the 80 Hz to 20 kHz from Erin's audiocorner to make better comparison. I like all the science based developments they advertise but I do not see the results in the MD Plots. Seems like in that scenario the SB Acoustics driver is better.

View attachment 457301
View attachment 457302

If you're comparing a Purifi vs other drivers, you might as well compare them in similar enclosures with similar crossover points. From the same designer as well.

Though I only linked to the IMD charts, both THD and IMD are ten or more dB lower on the Purifi driver throughout its entire range of operation. (Side note, it looks like a mislabeled chart for the C6B which should read 95 dB for the green line, then it would match the voltage)

 

[Heinrich]

If you're comparing a Purifi vs other drivers, you might as well compare them in similar enclosures with similar crossover points.

I feel that's not the point. See also my post

Sealed mini speaker, 3-way

Question is, were they successful? I can't really answer that. Anecdotally, I showed them at a meet and was quite surprised that they were very well received. That's why I asked which measurements should be made and for what purpose. What should they say in terms of preference by a single...

www.audiosciencereview.com

That little speaker was developed to showcase three- versus two-way. Just for the fun of it, and only by chance it went into use by somebody else. Yes, the experiment was set up only to debunk Purify. It is legal, you can do that. Result is as expected.

It was sparked by the demonstration found on Purify's site, linked above. My way bigger speakers made no difference between AM and Doppler aka FM. Are they, despite the effort in real estate, footprint, and regardless of the virtually perfect measurements that bad? I concluded that Doppler is a real limitation, that cannot be circumvented, but is mitigated by using n-ways, n as large as possible.

 

[amirm]

Another topic would be audilbility of MD? How should a level of MD be evaluated?

It has been studied. The highest levels of distortion are in bass. It has been shown that IMD products are at lower amplitude than harmonics and hence, doesn't reach audibility thresholds. See for example the AES Journal Paper:

Subwoofer Performance for Accurate Reproduction of Music*
LOUIS D. FIELDER Dolby Laboratories,
San Francisco, CA 94103, USA AND ERIC M. BENJAMI

"It will be shown that intermodulation distortion is equal to or less audible than harmonic distortion for the frequency range of subwoofer operation,"

[..]

"Thus intermodulation distortion that produces frequency difference products is not very important for subwoofer distortion analysis, since the masked thresholds will always be larger for signals below 100 Hz than above."

He found the same thing in determining audibility of distortion in headphones:

Perceptual Assessment of Headphone Distortion
Louis D Fielder Dolby Laboratories Inc.

Despite its title, the paper creates a distortion metric based on harmonics of a signal using single tones, not dual tone IMD. Analysis using single tones is far easier than dual tones which divide the non-linearities into smaller components, making them less audible and analysis far more complicated.

IMD test's value is that you don't have to capture beyond audio bandwidth. This advantage is of little value to us as we now have wideband measurement techniques. It also enables testing at two frequencies at once. Those two tones may not at all represent the worst case audibility metric. From above paper:

"Unfortunately, simple distortion metrics such as harmonic distortion, total harmonic distortion and noise, and intermodulation distortion do not correlate well with perceived quality."

 

[kharan]

I feel that's not the point. See also my post

Sealed mini speaker, 3-way

Question is, were they successful? I can't really answer that. Anecdotally, I showed them at a meet and was quite surprised that they were very well received. That's why I asked which measurements should be made and for what purpose. What should they say in terms of preference by a single...

www.audiosciencereview.com

My post was imho a relevant answer to Finn's post. It wasn't aimed at you.

That little speaker was developed to showcase three- versus two-way. Just for the fun of it, and only by chance it went into use by somebody else. Yes, the experiment was set up only to debunk Purify. It is legal, you can do that. Result is as expected.

It was sparked by the demonstration found on Purify's site, linked above. My way bigger speakers made no difference between AM and Doppler aka FM. Are they, despite the effort in real estate, footprint, and regardless of the virtually perfect measurements that bad? I concluded that Doppler is a real limitation, that cannot be circumvented, but is mitigated by using n-ways, n as large as possible.

3-ways produce much less IMD than two ways. I don't know what you are seeking to prove, it is already a known and well accepted fact.

 

[Finn]

If you're comparing a Purifi vs other drivers, you might as well compare them in similar enclosures with similar crossover points. From the same designer as well.

Though I only linked to the IMD charts, both THD and IMD are ten or more dB lower on the Purifi driver throughout its entire range of operation. (Side note, it looks like a mislabeled chart for the C6B which should read 95 dB for the green line, then it would match the voltage)

I actually wasn't aware the A6B did exist. it clearly is better. Question would be about the audibility of MD. Which is why I actually started this threat. And how to measure them without Klippel Software and Hardware.

 

[Heinrich]

My post was imho a relevant answer to Finn's post. It wasn't aimed at you.
3-ways produce much less IMD than two ways. I don't know what you are seeking to prove, it is already a known and well accepted fact.

Sure, now as we are told that IM is not as readily audible (in subwoofers), seems I ran out of fuel.

Only as an aftermath, the test/demo Purify provides on their site as linked above, no revelation? AM/IM is audible, and in case, Doppler is audible. What I wanted to prove is, that Purify is not correct, and the attempt to stretch excursion limits is limited by physics, to which even the auditory system obeys.

Thanks

 

[Finn]

It has been studied. The highest levels of distortion are in bass. It has been shown that IMD products are at lower amplitude than harmonics and hence, doesn't reach audibility thresholds. See for example the AES Journal Paper:

Subwoofer Performance for Accurate Reproduction of Music*
LOUIS D. FIELDER Dolby Laboratories,
San Francisco, CA 94103, USA AND ERIC M. BENJAMI

"It will be shown that intermodulation distortion is equal to or less audible than harmonic distortion for the frequency range of subwoofer operation,"

[..]

"Thus intermodulation distortion that produces frequency difference products is not very important for subwoofer distortion analysis, since the masked thresholds will always be larger for signals below 100 Hz than above."

He found the same thing in determining audibility of distortion in headphones:

Perceptual Assessment of Headphone Distortion
Louis D Fielder Dolby Laboratories Inc.

Despite its title, the paper creates a distortion metric based on harmonics of a signal using single tones, not dual tone IMD. Analysis using single tones is far easier than dual tones which divide the non-linearities into smaller components, making them less audible and analysis far more complicated.

IMD test's value is that you don't have to capture beyond audio bandwidth. This advantage is of little value to us as we now have wideband measurement techniques. It also enables testing at two frequencies at once. Those two tones may not at all represent the worst case audibility metric. From above paper:

"Unfortunately, simple distortion metrics such as harmonic distortion, total harmonic distortion and noise, and intermodulation distortion do not correlate well with perceived quality."

Well but this only is valuable for frequencis below 100 Hz, as stated if I understand correctly.

I don't think there is no correlation between distortion and perceived quality. If you read the KLIPPEL LIVE #14: Setting meaningful tolerances there are some thresholds defined for the audibility of different kinds of distortions. I will try to read the papers referenced.

 

[kharan]

Sure, now as we are told that IM is not as readily audible (in subwoofers), seems I ran out of fuel.

Only as an aftermath, the test/demo Purify provides on their site as linked above, no revelation? AM/IM is audible, and in case, Doppler is audible. What I wanted to prove is, that Purify is not correct, and the attempt to stretch excursion limits is limited by physics, to which even the auditory system obeys.

Thanks

I remember you saying that we should get vertical directivity as clean from lobing as as possible, even if the audibility of lobing in the vertical axis seems to be low. By the same reasoning, we should keep IMD as low as possible, even if it's audibility is dubious.

 

[Heinrich]

I remember you saying that we should get vertical directivity as clean from lobing as as possible, even if the audibility of lobing in the vertical axis seems to be low. By the same reasoning, we should keep IMD as low as possible, even if it's audibility is dubious.

Aha, what else to say, still no response to the Purify demo. You could, if proficient enough in that particular field, determine your personal threshold of audibility. I found it quite easy to do, and I'm quite normal. I came to being able to detect 0,3% in certain frequency bands, but accept 1% when listening to music. Klippel provides a self-test also, my threshold was < -53dB (~0,3%), but that's for sure just coincidence.

(On lobing, trying to take away my personal credibility, why is KEF succeeding with coax, whihch they long for sinde the 70s? Some people like coax, but why is that!?)

Oh, does the data you quoted comprise Doppler or not?

 

[kharan]

Aha, what else to say, still no response to the Purify demo. You could, if proficient enough in that particular field, determine your personal threshold of audibility. I found it quite easy to do, and I'm quite normal. I came to being able to detect 0,3% in certain frequency bands, but accept 1% when listening to music. Klippel provides a self-test also, my threshold was < -53dB (~0,3%), but that's for sure just coincidence.

(On lobing, trying to take away my personal credibility, why is KEF succeeding with coax, whihch they long for sinde the 70s? Some people like coax, but why is that!?)

Oh, does the data you quoted comprise Doppler or not?

I think it does.

 

[amirm]

I don't think there is no correlation between distortion and perceived quality.

There is of course correlation between harmonic distortion (NOT THD) and audibility. This analysis can be done and metrics developed as Fielder did. That is his specialty by the way and his papers are mandatory reading on these topics. As I stated, the weakness of IMD is that it creates a spray of distortion products, all of which are at lower amplitude than single harmonic spikes. That spray stays below threshold of hearing often when individual harmonics do not. This is why both Geddlee and Fielder's work are based on harmonic distortions and not IMD.

I don't know where the appeal of IMD is coming from. I just don't see that in research.