I understand there is a ton of info on this issue and it has been discussed debated argued etc., however I would like to hear what some here have to say about my specific experience. Basically it has to do with the audibility of 3rd harmonic distortion between about 300 and 1.5 KHz, and here is where I really don't even know what question to ask, are there differences in audibility due to different sources of said distortion??
Details..
This question arises from months of work developing two ribbon designs. These ribbons are to be used from 500 Hz on up. In one design the 2nd , 4th, and 5th harmonic are well below 0.5% with the 3rd harmonic rising to .35% at 1k and up to about .5% at 500 Hz. This at 90 db at 1 meter. Above about 1 KHz it reads lower but my focus is on the frequencies between about 300 to about 1.5 KHz. The other design (everything the same except a bit different diaphragm design) shows 2nd, 4th, and 5th well below .5% BUT the 3rd harmonic starts to rise below 1.5 kHz reaching approx. 1.5 % at 1k and about 2.3 % at 500 Hz. In both designs this rising 3rd is a smooth ramp up. Not a peaky looking curve.
So here’s the issue, once systems are all tailored to match each other in FR and levels, I/we cannot hear any significant difference between the two. Our listeners range from 26 years old to 58 yrs old. One has a lot of musician experience and sings ( 38 yrs old) .
So what’s my question? Well it would seem that we would simply choose the one that has lower measured distortion and be done with it. However for reasons I would rather not go into there is reason to want to develop the higher measured distortion unit further despite the higher distortion in this lower range. So just looking for your thoughts on this. I always hear people wax on and on about how “nasty” that 3rd harmonic can be and how anything above about .5% is not respectable etc etc. This latest development work seems to suggest there is more to the story and would love to start a war, um, I mean respectful conversation about why no one as yet has been able to hear anything ugly on our “high distortion” unit….
Details..
This question arises from months of work developing two ribbon designs. These ribbons are to be used from 500 Hz on up. In one design the 2nd , 4th, and 5th harmonic are well below 0.5% with the 3rd harmonic rising to .35% at 1k and up to about .5% at 500 Hz. This at 90 db at 1 meter. Above about 1 KHz it reads lower but my focus is on the frequencies between about 300 to about 1.5 KHz. The other design (everything the same except a bit different diaphragm design) shows 2nd, 4th, and 5th well below .5% BUT the 3rd harmonic starts to rise below 1.5 kHz reaching approx. 1.5 % at 1k and about 2.3 % at 500 Hz. In both designs this rising 3rd is a smooth ramp up. Not a peaky looking curve.
So here’s the issue, once systems are all tailored to match each other in FR and levels, I/we cannot hear any significant difference between the two. Our listeners range from 26 years old to 58 yrs old. One has a lot of musician experience and sings ( 38 yrs old) .
So what’s my question? Well it would seem that we would simply choose the one that has lower measured distortion and be done with it. However for reasons I would rather not go into there is reason to want to develop the higher measured distortion unit further despite the higher distortion in this lower range. So just looking for your thoughts on this. I always hear people wax on and on about how “nasty” that 3rd harmonic can be and how anything above about .5% is not respectable etc etc. This latest development work seems to suggest there is more to the story and would love to start a war, um, I mean respectful conversation about why no one as yet has been able to hear anything ugly on our “high distortion” unit….
Just one sample point but there was a web site with musical files for dl with injected second hd, and a seperate files with injected 3rd hd. Using it random test sequencing and blInd, in my set up with that music (piano), I could more easily detect 2nd HD and I found it more objectionable
I have built a lot of Class A amplifiers that have either 2nd or 3rd harmonic
distortion but not both, and gave them to listeners and/or sent them out
into the marketplace. I have received fairly consistent feedback which mostly
mirrors my own observations.
When you start getting above about .1%, people can hear the difference.
Not a new observation - in the '60's Julian Hirsch (much derided for his
supposed "can't hear it" stance) made a comment that he couldn't hear
distortion below .1%. He is to be praised for his honesty.
Interestingly, there seems to be a consistent subjective comment about
3rd, which is that it seems to make sound a little more dynamic, the
silences a little more black. 2nd harmonic is praised for warmth and/or
sweetness, but I have found it to also have two qualities about apparent
depth depending on the phase of the distortion, and I have commented on
this elsewhere (BAF 2017 on youtube for example).
I have noticed that some audiophiles like 2nd and some like 3rd, probably
for the character described above. Nobody seems to like harmonics above
2nd or 3rd, or at least won't admit to it... 😛
distortion but not both, and gave them to listeners and/or sent them out
into the marketplace. I have received fairly consistent feedback which mostly
mirrors my own observations.
When you start getting above about .1%, people can hear the difference.
Not a new observation - in the '60's Julian Hirsch (much derided for his
supposed "can't hear it" stance) made a comment that he couldn't hear
distortion below .1%. He is to be praised for his honesty.
Interestingly, there seems to be a consistent subjective comment about
3rd, which is that it seems to make sound a little more dynamic, the
silences a little more black. 2nd harmonic is praised for warmth and/or
sweetness, but I have found it to also have two qualities about apparent
depth depending on the phase of the distortion, and I have commented on
this elsewhere (BAF 2017 on youtube for example).
I have noticed that some audiophiles like 2nd and some like 3rd, probably
for the character described above. Nobody seems to like harmonics above
2nd or 3rd, or at least won't admit to it... 😛
hmmm interesting
One of the reasons I even mention this is the idea (yes I get in trouble here 😉) that there may be different audabilities depending on what is causing the distortion. For instance in another test we can take two absolutely identical ribbon designs with the only difference being about 1/4 mm difference in ribbon width and see a 100% or more difference in this 3rd harmonic below 1k. Two identical ribbons basically with just a very small change in the gap between ribbon and magnet face and watch the 3rd rise. BUT in listening there is no detectable difference between the two in listening. Now I am aware that the effects of that gap are complex and well beyond me but dang how is it that a change of that 3rd harm from about .5 to as much as 2% isnt heard?? I mean in some examples where we purposely use a diaphragm design that has noticeable degradation in sound quality and its immediately obvious, but a small change in gap that produces a "large" change in 3rd is not?? I just dont get that.
One of the reasons I even mention this is the idea (yes I get in trouble here 😉) that there may be different audabilities depending on what is causing the distortion. For instance in another test we can take two absolutely identical ribbon designs with the only difference being about 1/4 mm difference in ribbon width and see a 100% or more difference in this 3rd harmonic below 1k. Two identical ribbons basically with just a very small change in the gap between ribbon and magnet face and watch the 3rd rise. BUT in listening there is no detectable difference between the two in listening. Now I am aware that the effects of that gap are complex and well beyond me but dang how is it that a change of that 3rd harm from about .5 to as much as 2% isnt heard?? I mean in some examples where we purposely use a diaphragm design that has noticeable degradation in sound quality and its immediately obvious, but a small change in gap that produces a "large" change in 3rd is not?? I just dont get that.
According to Geddes, loudspeaker distortion (before things go off the rails) is basically inaudible even to quite high percentages.
He did his test by having a speaker in a room with a microphone, and that fed to headphones. The gain to the headphones was dropped 1dB for every dB that the drive to the speaker was increased. Apparently no one listening to the phones could tell what drive level was higher (more THD) or lower (less THD). I don't know what kind of music or how many nor what type listeners were tested.
Loudspeakers have some inherent acoustic filtering that makes them a different case from amplifiers, I'd think.
He did his test by having a speaker in a room with a microphone, and that fed to headphones. The gain to the headphones was dropped 1dB for every dB that the drive to the speaker was increased. Apparently no one listening to the phones could tell what drive level was higher (more THD) or lower (less THD). I don't know what kind of music or how many nor what type listeners were tested.
Loudspeakers have some inherent acoustic filtering that makes them a different case from amplifiers, I'd think.
From "Hearing, Its Psychology and Physiology" by the Acoustical Society of America, the audibility of 2nd harmonic distortion with a pure 370 Hz fundamental:
"The masking of an added harmonic is negligible below
a sensation level of 40 - 50 dB. From 50 to 80 dB,
the amount of harmonic necessary for an audible change
increases rapidly, first in absolute magnitude, and later
in relative magnitude as well...."
"...The qualitative character of the audible change produced
by adding this harmonic was different at the various sensation
levels of the fundamental. At low levels the harmonic was usually
heard as a separate tone. In the middle region [50 - 80 dBSPL]
it was heard as a sharpening or brightening of the timbre of the
tone, whereas at high levels the changes were so complex and so
dependent upon differences of phase that any generalization about
their character would be misleading."
The audibility of distortion depends upon the phase relationship between the harmonic and the fundamental since changes in this phase relationship will also change the musical envelope, and therefore the cochlear envelope in the inner ear. In effect, it allows the harmonic to become periodically unmasked and more audible. For example from "The representation of speech in the peripheral auditory system", there's a paper by Schroeder and Mehgart "Auditory Masking in the perception of speech", where they show that reversing the phase of even one harmonics component is audible. They could even "produce little melodies by sheer phase manipulations."
I recall a console designer claiming 7th harmonic was by far the worst since it was rarer in music. In the listening test I mentioned in my first post, I found 4th more obnoxious than even 5th (as high as the study went)
Frequency matters
It is relatively simple to generate a fundamental sine wave tone and a second tone at three times the first tones frequency to model third harmonic distortion. A program like Octave to do it with ease and you can listen to it with headphones. Depending on the frequency, the ear can be much more sensitive to the harmonic than it is to the fundamental. Pick a point on the Fletcher-Munson curve that is steep, 100 Hz. The ear is approximately 20 dB more sensitive to 300 Hz than it is to 100 Hz at low levels. So I suspect that 0.1% distortion of a 100 Hz signal may sound like 1% distortion due to the different sensitivity of the ear to those two different frequencies.
It is relatively simple to generate a fundamental sine wave tone and a second tone at three times the first tones frequency to model third harmonic distortion. A program like Octave to do it with ease and you can listen to it with headphones. Depending on the frequency, the ear can be much more sensitive to the harmonic than it is to the fundamental. Pick a point on the Fletcher-Munson curve that is steep, 100 Hz. The ear is approximately 20 dB more sensitive to 300 Hz than it is to 100 Hz at low levels. So I suspect that 0.1% distortion of a 100 Hz signal may sound like 1% distortion due to the different sensitivity of the ear to those two different frequencies.
To put it in to perspective, .5% distortion is about -45 dB from the fundamental, 2% is about -34 dB. At 90 dB, 2% is only 56 dB, a whisper level at a meter. With music playing at 90 dB, I wouldn't notice a whisper, and it is not even related to what is being played.
Back to the orders- second order is difficult to detect, because it does not change the musical composition at all, simply adding an upper octave harmonic. Odd orders can introduce changes which can add different notes, changing chord structure at very high levels. This can go from "interesting" to "disgusting" depending on the extent and the original chord structure.
Most "pop" music has such a large level of even and odd order distortion as recorded that adding another few percent makes little difference, though if the source is very clean and "sine wave like" to start (metal flute, for instance) a little distortion goes a long way. That said, distortion makes everything sound louder, and as Motörhead's immortal founder Lemmy said: "Make everything sound louder than everything else!"
Oy,
Art
Last edited:
Good advice. Several of the old audition tests from the 50s and 60s showed that distortion audibility followed equal loudness contours quite well
Lowpass> you mention the distortion below 1khz, but where were you crossing the ribbon at? I assume somewhat higher? That alone may be why you don't hear it, because it isn't being fed with enough @1khz signal to generate audible distortion. Or because even the 3rd harmonic for a ribbon at typical crossover (say 3.5khz) is pretty high in frequency (10.5khz) so may not be as audible. Just my .02
They are crossed over at 500 hz 2nd order LR so yes there is a little "cheating" here as the crossover is helping distortion figures a bit at 1k but I dont think its a large amount.
BTW there are two diaphragm designs . One shows this rising 3rd, the other is much better measurement wise anyway and shows about .6% at 500 and .4% at 1k at 90 db
Last edited:
The biggest issue IMHO is that IM distortion is really the elephant in the room.
We talk about harmonics partly as a way of isolating the problem, but the
same non-linearities rendered as IM in complex material are the larger effect,
and I think many of the perceived qualities are related to that, not so much
with results coming from pure/simple tones.
We talk about harmonics partly as a way of isolating the problem, but the
same non-linearities rendered as IM in complex material are the larger effect,
and I think many of the perceived qualities are related to that, not so much
with results coming from pure/simple tones.
Ear won’t mask IM the way it would HD. I’m partial to a multitone input covering the entire spectrum with powers emulating standards equivalent power spectral density of music, then reading the hash between the tones
I am reluctant to wade into this discussion as it has been beat to death elsewhere. First, I was quoted incorrectly so let me clear that up. Our study was for compression drivers only. We used a plane wave tube and headphones and adjusted the signals to playback always with the same level, but the DUT was played ever louder and louder. Different drivers were compared. What showed to be "significant" was frequency response, while Nonlinear distortion was not significant even at THD levels of 25%.
I also do not feel that talking about hearing a single harmonic level has much to do with how nonlinearities affect real musical signals. So I would take any tests like this with a grain-of-salt. Testing real signals through real systems is a far far more difficult test and one that virtually always results in entirely different results (to wit my own tests.)
I also do not feel that talking about hearing a single harmonic level has much to do with how nonlinearities affect real musical signals. So I would take any tests like this with a grain-of-salt. Testing real signals through real systems is a far far more difficult test and one that virtually always results in entirely different results (to wit my own tests.)
I put a demonstration of 2nd vs 3rd HD at 8:50 in this video using a guitar. You can hear the difference in musical terms.
YouTube
YouTube
- Status
- Not open for further replies.