Bob Cordell said:
Hi Bob, yes that would translate to 0.03% thd for the 10khz signal and 0.06% for the 20khz. Assuming that the distortion scales linear, an amp with 0.01% at 20khz would have at most +/- 1.66ns pim, or an amp with 0.001% a total of 0.33ns max.
Looking at the phase of the harmonics should tell the ratio pim to unlinearity.
I guess, the big question is still if pim is more audible than "normal" distortion. If not, we might have to look somewhere else for the reason of some amps sounding bad.
AFAIK, the ear is not interested in the phase shift of spectral content.
But on the other hand, what happens if the phaseshift of these harmonics is level dependent, wouldn't this smear up things as music contains tons of amplitude modulated signals ?
Mike
Michael,
I do not think it is very interesting. I made a lot of A-B tests with OPA627 x OPA134, e.g. People are easily able to tell the difference.
Do not restrict your point of view to THD, or another similar distortion. The opamp has 5 inputs. They are:
1) + input
2) - input
3) +Vs rail
4) -Vs rail
5) output
6) and possibly compensation pins.
These opamps work in agressive digital environment. They differ in PSRR, CMRR, noise, distortion, mix products. Why should they sound the same? And we argue by THD at full scale output voltage
I do not think it is very interesting. I made a lot of A-B tests with OPA627 x OPA134, e.g. People are easily able to tell the difference.
Do not restrict your point of view to THD, or another similar distortion. The opamp has 5 inputs. They are:
1) + input
2) - input
3) +Vs rail
4) -Vs rail
5) output
6) and possibly compensation pins.
These opamps work in agressive digital environment. They differ in PSRR, CMRR, noise, distortion, mix products. Why should they sound the same? And we argue by THD at full scale output voltage
MikeB said:
This is an old misconception rooted in our poor understanding of the cochlea workings, thinking of it as a more or less sensitive spectrum analyzer devoid of phase discrimination capability.
Recent work - as little back as year 2000 - have uncovered the stereocilia or hair cells responsible for signal transduction in the basilar membrane have an active role as amplifiers and quite nonlinear to be sure. This goes a long way to explain both sensitivity and Q, thinking in terms of active filter operation.
In the end and with respect to the phase issue, what matters is that the transduction in very nonlinear, meaning not only frequency information but *waveshape* make a perceptual difference.
Mechanics of the mammalian cochlea by Robles and Ruggero, Physiological Reviews, American Physiological Society, is an excellent review article on the subject.
Rodolfo
ingrast said:
Rodolfo,
I graduated in electroacoustics in 1979, i.e. 27 years ago. Even in the end of seventies, professor Merhaut, our lecturer, spoke about different mechanisms of human ear perception and about "waveshape" sensitivity.
Cheers,
Pavel
http://www.aes.org/aeshc/docs/jaes.obit/JAES_V52_12_PG1298.pdf
Bricolo said:
Yes, that's a loop back measurement. The card is m-audio delta 24b96. You get these low distortion levels only with reduced levels at 48khz. At full scale it looks like all other cards.
PMA said:
These opamps work in agressive digital environment. They differ in PSRR, CMRR, noise, distortion, mix products. Why should they sound the same? And we argue by THD at full scale output voltage
Thanks Pavel, good point ! I just prefer things to be measurable...
I just want to find out if PIM is interesting or just distortion.
As you already often pointed out, the DAC-residuals are very high frequency. (As an example)
ingrast said:
Rodolfo, thanks for pointing me to that. I was only aware of the perception of small phasehifts between left/right. (down to µSecs)
Mike
PMA said:
You were fortunate to attend lectures from such a referent !!
That waveshape had something to do in perception was long suspected, only it took sophisticated technology to confirm.
In this case, measurements of the active participation of stereocilia had to be performed "in vivo", painstakingly drilling tiny openings on the temporal bone and using laser interferometry to measure local basilar membrane vibrations. This is what was accomplished from year 2000 on.
Rodolfo
Audibility of PIM
John,
No question that the cheap IC is compromising the sound of your player. Its just that one does not need to even come close to discussing PIM to know that the op amp is garbage.
Although I think 1 ns of PIM is probably insignificant compared to other more conventional distortions that will inevitably accompany it, I'm not apologizing PIM away. My main point has been that going to wide open loop bandwidth makes little difference to PIM, and that the PIM contributed by negative feedback is almost always overshadowed by other sources of PIM.
So, if we all suddenly discovered that PIM of only 1 ns was indeed audible, I'd still be comfortable with the situation, since my focus is on what makes it smaller (and that my amplifier already has probably among the very lowest PIM).
Indeed, some people have ridiculed me for designing an amplifier with less than 0.001% THD out to 20 kHz at full power. But guess what? If PIM is more audible than we think, that means there was good reason for me to go for a much more linear amplifier, since all the evidence suggests that if you succeed in achieving ultra-low THD, you get ultra-low PIM. In other words, as a side benefit of my extremely linear design, I got PIM below 100 picoseconds.
I'm happy either way. I don't need to argue that PIM is relatively inaudible.
Bob
john curl said:
John,
No question that the cheap IC is compromising the sound of your player. Its just that one does not need to even come close to discussing PIM to know that the op amp is garbage.
Although I think 1 ns of PIM is probably insignificant compared to other more conventional distortions that will inevitably accompany it, I'm not apologizing PIM away. My main point has been that going to wide open loop bandwidth makes little difference to PIM, and that the PIM contributed by negative feedback is almost always overshadowed by other sources of PIM.
So, if we all suddenly discovered that PIM of only 1 ns was indeed audible, I'd still be comfortable with the situation, since my focus is on what makes it smaller (and that my amplifier already has probably among the very lowest PIM).
Indeed, some people have ridiculed me for designing an amplifier with less than 0.001% THD out to 20 kHz at full power. But guess what? If PIM is more audible than we think, that means there was good reason for me to go for a much more linear amplifier, since all the evidence suggests that if you succeed in achieving ultra-low THD, you get ultra-low PIM. In other words, as a side benefit of my extremely linear design, I got PIM below 100 picoseconds.
I'm happy either way. I don't need to argue that PIM is relatively inaudible.
Bob
traderbam said:
I understood him differently: he means that in the real life he prefers something different than in professional SPICE - simulated Matrix that is made of commonly shared beliefs.
I remember, back in 1991 one representative of big American corporation left the room of CFM of Russian automotive corporatin when he was asked: "Why you yourself don't use in your production computers you are trying to sell me?"
Hi, Bob Cordell,
In your opinion, what makes a non-global feedback sounds different than a global feedback amp? Non-global feedback amp (a good design) can sound more relaxed to our ears.
Usually it comes to output impedance (global feedback amp has low output impedance). But is this all the difference, or there is something else that differs from global feedback amp and non-global feedback amp, that our ear can hear?
How about this : a non global feedback amp, but has a Hawksford EC on the output stage (=low output impedance without global feedback). Will this sounds like global feedback amp or non-global feedback amp?
In your opinion, what makes a non-global feedback sounds different than a global feedback amp? Non-global feedback amp (a good design) can sound more relaxed to our ears.
Usually it comes to output impedance (global feedback amp has low output impedance). But is this all the difference, or there is something else that differs from global feedback amp and non-global feedback amp, that our ear can hear?
How about this : a non global feedback amp, but has a Hawksford EC on the output stage (=low output impedance without global feedback). Will this sounds like global feedback amp or non-global feedback amp?
MikeB said:
Yes, it seems that IC's do make a difference, perhaps for reasons that we don't understand. I'm not talking about PIM here. The 25-year-old NE5534 family of IC's has remarkably low distortion (including virtually unmeasurable PIM), yet there is a lot of evidence that newer op amps sound better. This, to me, remains a mystery of life in audio. I chalk things like this up to what I call the "X factor". These are things where there is pretty good evidence that there really is an audible difference, but where we have yet to have an objectively measurable hypothesis and measurement approach that will reveal a difference that could be responsible.
Bob
Audibility of Phase
I don't want to get us too far off topic, but it often seems that discussions somehow get around to whether or not the ear is sensitive to phase and waveshape. When we talk about this in the context of very small phase aberrations of electronic equipment, I can't help but wonder about the horrible things most loudspeakers do to phase, and think about the guy from the Bible with the log in his eye. Just look at the step response of any of the 99% of loudspeakers that are not transient-accurate. Go figure. My body shudders in the same way when people talk about time smear in any context other than loudspeakers. What am I missing here?
Bob
ingrast said:
I don't want to get us too far off topic, but it often seems that discussions somehow get around to whether or not the ear is sensitive to phase and waveshape. When we talk about this in the context of very small phase aberrations of electronic equipment, I can't help but wonder about the horrible things most loudspeakers do to phase, and think about the guy from the Bible with the log in his eye. Just look at the step response of any of the 99% of loudspeakers that are not transient-accurate. Go figure. My body shudders in the same way when people talk about time smear in any context other than loudspeakers. What am I missing here?
Bob
Re: Audibility of Phase
Agree we are wandering off topic, sorry.
True, reproduction systems are terrible in the end.
But then, why is it that reproduced (loudspeaker generated) sound is almost allways unmistakable recognized as such in comparison with the real thing?
We may "like" what we hear, but I still have not found something that, sounding behind a curtain, can fool me as to what is reproduced and what is original.
Rodolfo
Bob Cordell said:
Agree we are wandering off topic, sorry.
True, reproduction systems are terrible in the end.
But then, why is it that reproduced (loudspeaker generated) sound is almost allways unmistakable recognized as such in comparison with the real thing?
We may "like" what we hear, but I still have not found something that, sounding behind a curtain, can fool me as to what is reproduced and what is original.
Rodolfo