• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

THD has a much bigger effect on sound than you think

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
It is a speaker, and I think you know that its distortion linearly rises with level. I almost always give rather hints than pedantic analysis.

Pedantic analysis would be detailed and correct.
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
You can't because it doesn't happen unless the amp is clipping.



Of course you can hear 1200 Hz at 65dB when bass is at 90dB. You can even hear that when 1200 Hz is at 30dB. I am sure you can even hear a 2dB increase in 1200Hz at that point as well.

Here's the thing. The 1200Hz at these levels will be distortion free and close to noise levels. The 90dB bass is still easy to hear.
That bass will have its own harmonics. The speaker will generate its own harmonics as well. The amp will generate harmonics from the bass fundamental note itself + generate harmonics from the harmonics present in the music.
The bass being played has many harmonics from itself. The lower harmonics will be the highest, the higher the harmonic order the more those generated harmonics drop. At the 5th harmonic the contribution from the largest source (in your case the deep bass note) will not be there anymore it is in the noise floor below audibility levels. The amp does not contribute.
That bass note is only present very shortly at a high level. At that point it has a shitload of harmonics. The harmonics from that bass note are much higher in amplitude than the generated harmonics even by the ACA.
As I demonstrated the 'extra energy' from amp generated harmonics in the lower mids is just 0.026dB or so. You can't hear that.

I know what you want to prove, or think I know. When you play loud bass you hear the treble turn 'nasty'. This is not caused by the amp but is acoustic behavior of the speaker, headphone or room itself or an amp clipping or when you have a really nasty amp with tons of IMD you may be hearing that.
Maybe you are listening to a filterless DAC and the speaker or amp really doesn't like the produce >24kHz signals.. There can be plenty of reasons.

In any case... and amp with 0.1% distortion at max. power levels or even measured at 1W won't have 10% distortion in the treble.
The energy in the mids is lower by itself which could generate 2nd, 3rd, and 4th harmonics into the treble range. The bass cannot. Some upper bass harmonics maybe could but these are already down very low and the amp will add virtually no harmonics at lower levels.
IMD from the bass note (the fundamentals) will generate small amplitude poles right next to say a 1.2kHz note. Small and masked. You can't hear those. At least not the ones generated by the amp from the bass.

In the end an amp with 0.1% distortion at full power or at a certain level will not generate 10% at higher frequencies.
What you do see in Amirs plots is that distortion seems to rise in percentage when the signal level gets lower.
Here's the thing.... that rise is the noise floor. The smaller the signal the closer it is to the noise floor. That is what it shows.
Distortion does not increase at smaller levels but actually decreases.

Now we are talking about amps that do not have a crappy or poorly adjusted AB idle current. Those amps will have increased distortion when the level gets smaller. This too can be spotted in plots though and is not the case with modern amps and certainly not with the ACA.
Thanks for the details explanations.

Hmm, talking about audibility.....

I played around with the distort app and able to get just 5th harmonic of 0.1% (-60db) added to 250hz pure tone.
Here is the deltawave spectrum comparison. As you can see, the second file got that 0.1% 5th harmonics.
250_tone_5th_harmonic_60db_down_deltawave_orig_spectrum.PNG


I tried to ABX pure tone vs file with that 5th harmonic. Very difficult, but still able to get 8/8.
250_tone_5th_harmonic_60db_down_8_0f_8_abx.PNG
 

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
So, subtract 15db for the crest factor and 20-30db for treble being that much quieter. You're at -35 to -45db now. If harmonic distortion is at -60, it's 15-25db down. Audible, easily audible

No.

You are trying to use a bunch of averages w.r.t crest factor and average ratio of bass to treble energy (should be mid band) and then makes the wild logical leap that that applies directly to a singular event in time.

It should be quite obvious it does not.

You are assuming that instantaneous treble tones 35 db down from a bass town are instantly evident (distortion is temperally colocated) and making the further assumption that something harmonically related another 15-20dd below that is detectable in music.

Let's kàeep the leaps of faith to the church. You statement of easily audible is just a leap of faith.
 

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
Thanks for the details explanations.

Hmm, talking about audibility.....

I played around with the distort app and able to get just 5th harmonic of 0.1% (-60db) from 250hz.
Here is the deltawave spectrum comparison. As you can see, the second file got that 0.1% 5th harmonics.
View attachment 160828

I tried to ABX it. Very difficult, but still able to get 8/8.
View attachment 160829

Now place it in music.
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
Now place it in music.
Sorry, for now, I don't really trust the distort app with music file. The distort app seems to have bugs. Using 250hz tone, I could get distortion entered to show up correctly in saved file. However, using 1khz pure tone, the saved file is showing like 10db less distortions levels than as entered in the app.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
@Pdxwayne

A: That's a single tone ? not music which would have signals above the 1250 level and near it so will be masked by the music.
B: That's a huge 5th harmonic only and no amp in the real world would show this kind of behavior.
C: That signal is what comes out of the DAC it is not what comes out of the speaker.
D: ACA at 2 Watts is -85dB 5th harm. Simulate that with both tones and music.

You are essentially playing a 250Hz tone + 1250 Hz tone at a lower level (and phase related). Of course you will hear that.
 

audio2design

Major Contributor
Joined
Nov 29, 2020
Messages
1,769
Likes
1,830
Sorry, for now, I don't really trust the distort app with music file. The distort app seems to have bugs. Using 250hz tone, I could get distortion entered to show up correctly in saved file. However, using 1khz pure tone, the saved file is showing like 10db less distortions levels than as entered in the app.

Don't start with the fundamental at 0db. Too much potential for math errors and intersample overs on playback.
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
@Pdxwayne

A: That's a single tone ? not music which would have signals above the 1250 level and near it so will be masked by the music.
B: That's a huge 5th harmonic only and no amp in the real world would show this kind of behavior.
C: That signal is what comes out of the DAC it is not what comes out of the speaker.
D: ACA at 2 Watts is -85dB 5th harm. Simulate that with both tones and music.

You are essentially playing a 250Hz tone + 1250 Hz tone at a lower level (and phase related). Of course you will hear that.

A. Music can briefly has single tone too.
B. Indeed, unlikely, just to show you masking of tone might not always happens.
C. Don't understand.
D. Indeed. Just a test to show masking as you mentioned, might not always true.

Yeah, you try ABX with 1250 hz at 60db lower than 250hz vs pure 250hz.
: P
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
A. Music can briefly has single tone too.

It will have harmonics by itself and there won't be a bass at high levels.
C. Don't understand.
The signal you showed will come out of the DAC.
What comes out of the speaker will certainly NOT look like what comes out of the DAC is my point. Yes both frequencies will be there but also lots of added harmonics by the speakers/headphones.
D. Indeed. Just a test to show masking as you mentioned, might not always true.

You show this by using a 2 tone continuous test signal with levels that won't happen in reality (when caused by an amp) ?
You showed that you can hear the difference between a single pure sine and a two tone pure continuous sine much louder than would occur in reality.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Yeah, you try ABX with 1250 hz at 60db lower than 250hz vs pure 250hz.

It is logical everyone can hear this.
It becomes a totally different matter when using actual music from an amp that only has 0.1% 5th harmonic ONLY distortion (which does not exist) at levels well below clipping levels.

Here's an average poor performing amp
index.php

It has a shitload of equally loud not harmonically related frequencies of which the total energy is 5W in 4 ohm.
In the sensitive area (around 3kHz) we see some harmonics (incl IM) that are over 90dB lower.

Here the amp with much higher output level (so higher distortion) where the higher harmonics (OP fears 3rd the most it seems) are -95dB lower.
For you to hear that there would only be a single bass playing at 100dB SPL and you might hear some faint higher frequency 'noise' going up only at the moment the bass is played. Your speaker will have much, much higher amounts of distortion, you may even hear speaker port noise much much louder.
It's a non issue with music. You have to remember that notes decay. when they do the harmonics also go lower into the noise floor of the amp and room.

index.php
 
Last edited:

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
.....
The signal you showed will come out of the DAC.
What comes out of the speaker will certainly NOT look like what comes out of the DAC is my point. Yes both frequencies will be there but also lots f added harmonics by the speakers.

.....
.
I see. I listened at moderate volume. E30 to L30 to Akg 371. Any extra harmonics I would hear would be mostly the same for pure 250hz vs the dual tones. So, that should not matter at all for abx
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
It is logical everyone can hear this.
It becomes a totally different matter when using actual music from an amp that only has 0.1% 5th harmonic ONLY distortion (which does not exist) at levels well below clipping levels.
Didn't you lectured me about masking? Are you sure everyone can hear 1250hz when 250hz is 60db louder?

You can prove to me right now. Do the same ABX and show the results here.
: P
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Your test has nothing to do with masking. There is no masking here. There are 2 continuous tones far enough apart that masking does not occur with single continuous tones.
As long as the lower frequency is louder than 90dB SPL it is easy to hear.
No amp will do this (when not broken), not with tones not with music it is a pointless test !!
Besides the transducers will add their own garbage exceeding that of what your source generated.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I see. I listened at moderate volume. E30 to L30 to Akg 371. Any extra harmonics I would hear would be mostly the same for pure 250hz vs the dual tones. So, that should not matter at all for abx

Let's agree that the K371 is generating distortion mostly lower order. Lets agree that you are using continuous tones. Lets agree the tones are far apart to not be bothered by masking. Lets agree that you add a signal to the already present distortion of the headphone.
Even at 94dB SPL the 3rd harmonic at 1.2kHz of the K371 was measured at 84dB distance. 5th will be lower.
This means you ADD an audible amount of continuous tones (at least 30dB higher than the harmonic the K371 would produce) that by reason alone should be audible.

Again ... this is NOT music
Your test does NOT prove what you want to prove.
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
Your test has nothing to do with masking. There is no masking here. There are 2 continuous tones far enough apart that masking does not occur with single continuous tones.
As long as the lower frequency is louder than 90dB SPL it is easy to hear.
No amp will do this (when not broken), not with tones not with music it is a pointless test !!
Besides the transducers will add their own garbage exceeding that of what your source generated.
A louder signal can mask a quieter signal, don't you agree? If I talk to you normally in front of you at over 90db, can you hear another person singing at 600 meters away? Isn't that call masking?

OK. Let's not use 5th harmonics as example. How about 3rd harmonics? More realistic?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
If I talk to you normally in front of you at over 90db, can you hear another person singing at 600 meters away? Isn't that call masking?

Some parts of the singer will certainly be masked by your voice. You just don't realize it because the voice is dominant. The moments the voice is quiet you will hear the singer.
It would be another matter if the singer did not sing but said the exact same words as you in the same tone and a bit earlier to compensate for the time delay of 600m. Then you won't hear that second person... THAT is masking.
When that person sneezes and the person close to you doesn't than that sound won't be masked unless you sneeze as well.

You need to read up on masking not theorize away on plots that have no relevance and doing absolutely pointless self-tests at unrealistic levels.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,940
Location
Brussels, Belgium
Absolutely. Harmonic distortion of 3% is normal in speakers at a medium or higher volume. Harmonic distortion automatically creates intermodulation distortion with difference tone of about same %.
Also, it is normal in many tube amplifiers and again this will create this amount of intermodulation distortion. Yes we all know that in systems with transfer function like speakers or tubes the distortion rises with signal amplitude and it will be lower for small signals and quiet music.

i think it's much more audible when you're listening to it this way (on headphones with low distortion/low excrsion at low volume) than a particular peak on a track on speakers (high volume thus higher masking and very transiently).

yes 3% is reasonable for a lot of bookshelves but not necessarily all the time..
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
never unless it's synth or done with a tone generator.
Indeed, I was thinking about electronica type of music. I used one of those music clip with simple single tone for abx before.
 

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,422
Likes
2,407
Location
Sweden
One could also use the human whistle.
 
Top Bottom