Low distortion in AVR vs sound quality

[lashto]

...
AES tests established 0.01% as the solid value for audio transparency.
...

Did a giant lot of Distortion reading lately but did not find any "solid value" anywhere. A link to those AES tests would also be highly appreciated.

 

[RichB]

In my opinion that is more so the case on Audioholics.com. On ASR, it seems more like a 50/50 split.

Forums will be forums. I like AH but there is very little new data there. ASR on the other hand is expanding the knowledge base.

- Rich

 

[Foxxy]

Did a giant lot of Distortion reading lately but did not find any "solid value" anywhere. A link to those AES tests would also be highly appreciated.

The best I could find is Fielder and Benjamin's paper where they focus on bass:
https://www.aes.org/e-lib/online/browse.cfm?elib=5147
You can see a little snippet here:
https://www.audioholics.com/loudspe...tortion-at-bass/total-harmonic-distortion-thd
I simply can't find the paper/AES journal article where this was done for non-bass frequencies but it was much older than '88.

I'd love to be able to contact NWAvGuy, I think he read it and maybe would remember but he's gone

So yeah, I can't serve up more proof than my memory, especially since all modern papers deal with PEAQ and the question "How to best compress audio data while still sounding good".

 

[mhardy6647]

In my opinion that is more so the case on Audioholics.com. On ASR, it seems more like a 50/50 split.

I think the weighting is towards by the numbers here than a random chance (50/50) distribution... but that's OK. It's a nice counterpoint to most of the rest of the hifi audio discussion presence on teh webz.

 

[peng]

I think the weighting is towards by the numbers here than a random chance (50/50) distribution... but that's OK. It's a nice counterpoint to most of the rest of the hifi audio discussion presence on teh webz.

Just rough number, never did any real counts lol!!

 

[mhardy6647]

Just rough number, never did any real counts lol!!

So what you're saying is that we should consult Car Talk's resident statistician Marge Innovera.
I believe she may be taking new clients
https://www.cartalk.com/content/staff-credits

 

[lashto]

The best I could find is Fielder and Benjamin's paper where they focus on bass:
https://www.aes.org/e-lib/online/browse.cfm?elib=5147
You can see a little snippet here:
https://www.audioholics.com/loudspe...tortion-at-bass/total-harmonic-distortion-thd
I simply can't find the paper/AES journal article where this was done for non-bass frequencies but it was much older than '88.

I'd love to be able to contact NWAvGuy, I think he read it and maybe would remember but he's gone

So yeah, I can't serve up more proof than my memory, especially since all modern papers deal with PEAQ and the question "How to best compress audio data while still sounding good".

Thanks for the update, I knew that one. It only applies to bass freq where our ears have the least D sensitivity. By quite a wide margin.
And btw, the whole Human Hearing series on AH is also very good (and pretty much everything else by the same author).

 

[Lbstyling]

Did a giant lot of Distortion reading lately but did not find any "solid value" anywhere. A link to those AES tests would also be highly appreciated.

I found a reference to Toole stating that distortion is audiable at 0.001%THD, and that basically generally lower distortion is preferential (other studies have illuded to this)

Quote from Toole on Audioholics:

"As for monitoring waveforms on an oscilloscope, I did that first about 50 years ago when I began to look at loudspeaker performance. When I measured harmonic distortion that could be heard in pure tones, the effect on the waveform shape was often simply not visible. But, back then a few pure class B power amps existed. With a pure tone the crossover distortion was easily audible, THD measured about 0.001%, normally thought of as negligible. The tiny blip in the waveform at the zero-voltage transition between the push and pull sections could be seen in the waveform at great magnification, but when the energy was averaged over a whole period of the waveform, it was vanishingly small. So, it became evident that distortion mechanisms that were locked to the waveform might be very differently perceived than those that were long-term averaged over the test signal duration. So, any measurement relying on RMS, energy-averaged, values are telling only part of the story. Was the crossover distortion audible in music? Not that anyone could tell, even with oboe solos in a reverberant concert hall. But that does not excuse it.

ADDITION One of the most difficult forms of distortion to measure, but one of the easiest to hear is what is called "rub and buzz" in which the voice coil is rubbing or some grit is in the gap around it. It generates tiny spikes of energy, correlated with cone movements - i.e. waveforms - but hardly shows up in energy-averaged distortion measurements.

As I said in my last post, the distortion one measures is totally related to the test signal used to probe the non-linearity. With music and movies that is infinitely variable.

The THD measurement was left in place because it happened almost automatically and a few loudspeakers would exhibit narrow-band distortion spikes when they misbehaved. It was left in the test repertoire for that reason only, not because this simple metic has any great meaning. We need a metric that can eliminate or attenuate distortion components that are perceptually masked. Then the numbers have a chance of telling us something useful. In the meantime, the closer to zero distortion, the better."

Statements seem balanced and logical to me.

There is something going on here though that has passed me by before. I would agree that there is no way you can hear a distortion at -80db that is playing 'in parallel' with the music notes at 0db(am I correctly describing linear distortion here and the masking effect?

But how do you describe a distortion that plays during the time gaps in time between notes? Because presumably I would be saying that these would need to be below the room noise floor to be inaudible.

When THD is calculated, if the volume of this distortion is calculated as a percentage of the peak SPL output, that is indeed a pointless way of describing the distortion IMO.

 

[lashto]

I found a reference to Toole stating that distortion is audiable at 0.001%THD, and that basically generally lower distortion is preferential (other studies have illuded to this)

Quote from Toole on Audioholics:

"As for monitoring waveforms on an oscilloscope, I did that first about 50 years ago when I began to look at loudspeaker performance. When I measured harmonic distortion that could be heard in pure tones, the effect on the waveform shape was often simply not visible. But, back then a few pure class B power amps existed. With a pure tone the crossover distortion was easily audible, THD measured about 0.001%, normally thought of as negligible. The tiny blip in the waveform at the zero-voltage transition between the push and pull sections could be seen in the waveform at great magnification, but when the energy was averaged over a whole period of the waveform, it was vanishingly small. So, it became evident that distortion mechanisms that were locked to the waveform might be very differently perceived than those that were long-term averaged over the test signal duration. So, any measurement relying on RMS, energy-averaged, values are telling only part of the story. Was the crossover distortion audible in music? Not that anyone could tell, even with oboe solos in a reverberant concert hall. But that does not excuse it.

ADDITION One of the most difficult forms of distortion to measure, but one of the easiest to hear is what is called "rub and buzz" in which the voice coil is rubbing or some grit is in the gap around it. It generates tiny spikes of energy, correlated with cone movements - i.e. waveforms - but hardly shows up in energy-averaged distortion measurements.

As I said in my last post, the distortion one measures is totally related to the test signal used to probe the non-linearity. With music and movies that is infinitely variable.

The THD measurement was left in place because it happened almost automatically and a few loudspeakers would exhibit narrow-band distortion spikes when they misbehaved. It was left in the test repertoire for that reason only, not because this simple metic has any great meaning. We need a metric that can eliminate or attenuate distortion components that are perceptually masked. Then the numbers have a chance of telling us something useful. In the meantime, the closer to zero distortion, the better."

statements seam reasonable and logical to me.

It's all true and logic and useful.
But it only says that 0.001% is "normally thought of as negligible". For me that sounds quite far from "inaudible". Pretty good though as a rule of thumb (at least as of 2020).

 

[Lbstyling]

It's all true and logic and useful.
But it only says that 0.001% is "normally thought of as negligible". For me that is (still) quite far from "inaudible". Pretty good though as a rule of thumb (at least as of 2020).

I was already in the camp that distortion mattered more than many would say, but reading this combined with the comments in Toole's last book on the ear being sensitive to comparitively small driver resonance peaks (hard material breakup) is a big leap for me.

This is what i found using the SEAS excel magnesium drivers in my first speakers. They measure exceptionally well.......but.....what Amir found with the bambosa speaker is the same as I found with my 3 way.....even though the crossover filter should put the distortion way way down, you can hear somthing wrong IMO.

It's also worth noting that when Toole says 'negligible', (or anything else for that matter) he is writing on a world of pre-Toole or anti-Toole retoric, and probably more to the masses of audio equipment like Samsung and Sony regarding TVs and now smart speakers rather than the very 'niche' world of audiophiles.

As a side note, many of the best smart speakers have taken Toole's words very seriously, to the extreme even when you look at the top Alexa and Siri devices.

 

[lashto]

I was already in the camp that distortion mattered more than many would say, but reading this combined with the comments in Toole's last book on the ear being sensitive to comparitively small driver resonance peaks (hard material breakup) is a big leap for me.

This is what i found using the SEAS excel magnesium drivers in my first speakers. They measure exceptionally well.......but.....what Amir found with the bambosa speaker is the same as I found with my 3 way.....even though the crossover filter should put the distortion way way down, you can hear somthing wrong IMO.

It might take a while until speaker drivers get anywhere close to 0.001% (-100dB). I don't have much hope for the next 10-20 years, but you never know, breakthroughs can happen anytime.

And btw, here is the latest DBT to show that -75dB THD is audible. And that was the 'friendliest' form of HD made mostly of H2/H3. Current speakers drivers with HD spikes in the ~50dB area aren't even close to that.

 

[Aerith Gainsborough]

Wanna know how 10%TD (total distortion, THD+IMD) sounds? In the worst possible makeup? Completely disharmonic?
http://ethanwiner.com/audibility.html

Have fun, see how subtle you can hear and if you have multiple audio playback devices, see on which you can go the lowest

Ouch, that sounds even nastier than the Klippel test.

I could hear it clearly in the Concerto 60 sample but was unable to detect it in the Concerto 70 sample.
I could hear the tones in the background of the tones-20 file but not the tones 30 one.
Speech was inaudible in the speech 20 file.
Music tones 40 was clear, music tones 50 inaudible.

As I thought: I am fine with an AVRs harmonics being @ -85dB.

 

[Lbstyling]

It might take a while until speaker drivers get anywhere close to 0.001% (-100dB). I don't have much hope for the next 10-20 years, but you never know, breakthroughs can happen anytime.

And btw, here is the latest DBT to show that -75dB THD is audible. And that was the 'friendliest' form of HD made mostly of H2/H3. Current speakers drivers with HD spikes in the ~50dB area aren't even close to that.

I'm at or close to this with a JBL 2452SL .....yay!
DIY though.

For a 'regular' shape speaker, I think the new Belisima tweeter and the SB acoustics CAC 6 inch or new purefi 6.5 inch bass drivers are about the best currently available, but yes the Belisima is -50-60db for 2nd harmonics then -75db for 3rd. You would need multiple bass drivers to get lower distortion with say 100db SPL at 1m peak regardless.

If you go over to 1.5 inch exit horns, you can go to around -70db, and the best of the big 15 inch drivers like the AE TD15m are around -60db at 95db/1m /1w. Have 4 of these and you could probably get around -75db distortion at 92db+

https://hificompass.com/en/reviews/bliesma-t34b-4-beryllium-dome-tweeter

 

[lashto]

I'm at or close to this with a JBL 2452SL .....yay!
DIY though.

For a 'regular' shape speaker, I think the new Belisima tweeter and the SB acoustics CAC 6 inch or new purefi 6.5 inch bass drivers are about the best currently available, but yes the Belisima is -50-60db for 2nd harmonics then -75db for 3rd. You would need multiple bass drivers to get lower distortion with say 100db SPL at 1m peak regardless.

If you go over to 1.5 inch exit horns, you can go to around -70db, and the best of the big 15 inch drivers like the AE TD15m are around -60db at 95db/1m /1w. Have 4 of these and you could probably get around -75db distortion at 92db+

https://hificompass.com/en/reviews/bliesma-t34b-4-beryllium-dome-tweeter

I don't know of any speaker driver that even approaches -75dB THD (or at the least has the H2 at ~that level). Those Bliesma look interesting, I guess very expensive too.

 

[Lbstyling]

I don't know of any speaker driver that even approaches -75dB THD (or at the least has the H2 at ~that level). Those Bliesma look interesting, I guess very expensive too.

The JBL Everest would be getting close, at least in the treble and mids.

 

[Flak]

I'm not sure that a listener will identify a 10% distortion, if present for a few milliseconds only, as happens when clipping...
without going to the extreme maybe relatively high percentages of distortion are not audible at all if the duration is short during the reproduction of peaks of a music track?

 

[RichB]

I'm not sure that a listener will identify a 10% distortion, if present for a few milliseconds only, as happens when clipping...
without going to the extreme maybe relatively high percentages of distortion are not audible at all if the duration is short during the reproduction of peaks of a music track?

I think it is well established that the nature of the distortion, frequency and duration, is deterministic.
At low levels it may be possible to hear distortion and noise as well.
IMO, distortion should be eliminated where possible in electronics and processing as its character is different from speaker distortion.

Distortion from subwoofers is generally considered very forgiving but higher order products can increase localization.

- Rich

 

[Flak]

I think it is well established that the nature of the distortion, frequency and duration, is deterministic.
At low levels it may be possible to hear distortion and noise as well.

Agreed, but with reference to AVRs as discussed here, I expect that distortion will be lower at low levels while it seems to me that we are measuring it at appx. full output
Also I didn't read the full DBT about audibility of distortion mentioned by lashto so I ask him... has distortion been added during the full length of the reproduction?
If so I wonder how close is it to the real world where I think that higher distortion is present only for a VERY short time

P.S. Mine are questions... not answers

 

[lashto]

Agreed, but with reference to AVRs as discussed here, I expect that distortion will be lower at low levels while it seems to me that we are measuring it at appx. full output
Also I didn't read the full DBT about audibility of distortion mentioned by lashto so I ask him... has distortion been added during the full length of the reproduction?
If so I wonder how close is it to the real world where I think that higher distortion is present only for a VERY short time

P.S. Mine are questions... not answers

The 75dB HD in that test was added in the file so yes it's same across all freqs and volumes.
Generally you are also right that the HD products are lower at lower volume. But the noise floor is also higher at lower volume so the whole D picture might be worse. On short: it's complicated

 

[Flak]

The 75dB HD in that test was added in the file so yes it's same across all freqs and volumes.
Generally you are also right that the HD products are lower at lower volume. But the noise floor is also higher at lower volume so the whole D picture might be worse. On short: it's complicated

Agreed, as you say noise floor is important at low levels... but this reasoning leads to thinking that THD+N measurements at full output are much less relevant than usually thought (as transient peaks in music last for milliseconds and even 10% distortion might be inaudible) while measurements at low level, on the contrary, are probably more important.
Your thoughts?
Am I missing something?