andy_c
I was just wondering if it would be worth considering the phase of the third harmonic, w.r.t. the fundamental.
I was just wondering if it would be worth considering the phase of the third harmonic, w.r.t. the fundamental.
Swedish Chef said:
So, MikeB - what is your current standpoint on phaseshifts? 😕😉
/M
Very simple !
Phaseshifts between left and right channel are VERY important, the
brain uses theses for locating sounds left/right. The volumedifference
is nearly negligible. I found out this when writing 3d-positional sound.
Only applying correct phaseshifts enabled precise left/right locating.
(Able to locate position down to ~10°)
Even more fascinating: The eardist has to be correct, or the effect
vanishes. This means that extreme phaseshifting can destroy the
stereoimaging.
Locating front/behind and up/down are much more complex, these
depend on the form of the ear, reflections from shoulder and such stuff...
Try it, take a monosignal, play it on the left channel 0.5ms delayed
to the right, you will have the feeling that the sound is right of you !
The dynamic phaseshifts introduced by an amp should be too small
to have a big impact on stereoimaging.
Mike
Check this: http://www.aip.org/pt/nov99/locsound.html
Interaural phase difference sensitivity in range of 13 microseconds. Interesting dependance on the frequency.
Interaural phase difference sensitivity in range of 13 microseconds. Interesting dependance on the frequency.
Sorry for introducing a different type of phasehift. The discussion
began with Phasemodulation, it's something completely different
than dynamic groupdelay.
Now some "believing":
I think mikeks is right. My experiments with phaseshifts convinced me,
that human ear is only interested in freqspectrum, means only distortions
creating harmonics are audible. (Except the stereothing)
But if i understood this PM-stuff, it means that when an amp is playing
complex signals it begins to generate a hell of thd, even if it can perfectly
reproduce a sinewave at any freq.
In sims i fed my amp with a multisignal (3khz+4khz). Suddenly it showed
harmonics at 1khz and 7khz. None of the 2 freqs should be able to
generate these. Is it correct that this is the modulation you are
talking about ?
Mike
began with Phasemodulation, it's something completely different
than dynamic groupdelay.
Now some "believing":
I think mikeks is right. My experiments with phaseshifts convinced me,
that human ear is only interested in freqspectrum, means only distortions
creating harmonics are audible. (Except the stereothing)
But if i understood this PM-stuff, it means that when an amp is playing
complex signals it begins to generate a hell of thd, even if it can perfectly
reproduce a sinewave at any freq.
In sims i fed my amp with a multisignal (3khz+4khz). Suddenly it showed
harmonics at 1khz and 7khz. None of the 2 freqs should be able to
generate these. Is it correct that this is the modulation you are
talking about ?
Mike
MikeB, this is simple IMD (intermodulation) distortion where you get "mixing products" at 4+3=7 kHz and 4-3=1 kHz. Nothing spectacular going on here. 😉
/Magnus
Mister John Curl
This discussion should be about technical issues regarding PIM, TIM and what not. Like the excellent arguments Andy_C is providing. This is 2004 and not 1974 (when I was not even born yet) so clearly the discussion should center around what we know now and what capabilities we have today and not what we knew 30 years ago.
There have been numerous posts here showing that any occurance of PIM or TIM easily can be detected and/or avoided. Every time you respond with some 70's tale and a personal reflection of the sonic marvels of state-of-the-art opamps like the uA709.
While they might be entertaining they add nothing to the technical discussion.
Maybe you get upset discussing these matters as a result of a giant food-fight back in the days. Neither do I know anything about it nor do I care.
Can we please proceed with this discussion as grown-up adults?
Regards
/Magnus Samuelsson
This discussion should be about technical issues regarding PIM, TIM and what not. Like the excellent arguments Andy_C is providing. This is 2004 and not 1974 (when I was not even born yet) so clearly the discussion should center around what we know now and what capabilities we have today and not what we knew 30 years ago.
There have been numerous posts here showing that any occurance of PIM or TIM easily can be detected and/or avoided. Every time you respond with some 70's tale and a personal reflection of the sonic marvels of state-of-the-art opamps like the uA709.
While they might be entertaining they add nothing to the technical discussion.
Maybe you get upset discussing these matters as a result of a giant food-fight back in the days. Neither do I know anything about it nor do I care.
Can we please proceed with this discussion as grown-up adults?
Regards
/Magnus Samuelsson
john curl said:
All very interesting John, but we now know that there is no justification for ridding ourselves of negative feedback on the false grounds that PIM is aggravated by it's introduction....
You made me feel about a million years old.
Magnus, sometimes the historical background can help to clarify issues and help place them into proper context. And since John continues to crank out well-regarded amps (even in 2004!), I think he's worth paying attention to, even if he's not speaking directly to your particular questions.
The issue is that the real life shortcomings have been dissolved into distant academic discussion here, IMHO. Individuals supporting their ego's, nothing more.
Well, at least we have had an interesting intellectual 'foodfight'. 😀
Well, this is how I see it, from a historical perspective:
"Those who chose to ignore history, will live to repeat it."
We seem to understand PIM in op amp designs about as well as we understood TIM (30 years ago). What this means is that we are 'stumbling' around trying to understand something that might help us make better audio products.
Let me 'again' give a brief history to TIM to clarify this.
1966, D&G IEEE Trans Audio: Qualitative description of TIM and first recommendation of high open-loop bandwith
1970 MO IEEE Trans Audio: Circuit examples TIM, still qualitative. 1966, D&G article footnoted.
1973 MO & JL at Phillips Research: Power amp design example with low TIM Excellent amp, I bought the prototype and used it for 15 years. Have 2'nd amp of same design in my lab audio system today. Something is 'right sounding' about this amp.
1976 MO, L & JC at Finnish Gvt Lab: Quantitative measurements of TIM and creation of standardized test.
1978 JC at own lab: Measurements of what would cause TIM in real audio sources.
ETC
Where are we with PIM today, compared to TIM? I would say about: 1970.
1980 MO PIM AES paper given in Europe. Qualitative only
1998 BG PIM paper discussed at this time. Quantitative examples.
Successful designers actually LISTEN to their creations and those of others. This provides them with the 'feedback' necessary to see if they are on the right track. Many of us have tried the 'high feedback', multiloop, and other approaches, with some disappointment. That is why we tend toward lower feedback and higher open loop bandwidth. Is it because of PIM? We don't know yet, and neither does anyone else, BUT we know what sounds good, and you will find examples of our designs in Class A listening ranking. We don't get that designation by NOT listening to our design efforts and modifying them accordingly.
Well, this is how I see it, from a historical perspective:
"Those who chose to ignore history, will live to repeat it."
We seem to understand PIM in op amp designs about as well as we understood TIM (30 years ago). What this means is that we are 'stumbling' around trying to understand something that might help us make better audio products.
Let me 'again' give a brief history to TIM to clarify this.
1966, D&G IEEE Trans Audio: Qualitative description of TIM and first recommendation of high open-loop bandwith
1970 MO IEEE Trans Audio: Circuit examples TIM, still qualitative. 1966, D&G article footnoted.
1973 MO & JL at Phillips Research: Power amp design example with low TIM Excellent amp, I bought the prototype and used it for 15 years. Have 2'nd amp of same design in my lab audio system today. Something is 'right sounding' about this amp.
1976 MO, L & JC at Finnish Gvt Lab: Quantitative measurements of TIM and creation of standardized test.
1978 JC at own lab: Measurements of what would cause TIM in real audio sources.
ETC
Where are we with PIM today, compared to TIM? I would say about: 1970.
1980 MO PIM AES paper given in Europe. Qualitative only
1998 BG PIM paper discussed at this time. Quantitative examples.
Successful designers actually LISTEN to their creations and those of others. This provides them with the 'feedback' necessary to see if they are on the right track. Many of us have tried the 'high feedback', multiloop, and other approaches, with some disappointment. That is why we tend toward lower feedback and higher open loop bandwidth. Is it because of PIM? We don't know yet, and neither does anyone else, BUT we know what sounds good, and you will find examples of our designs in Class A listening ranking. We don't get that designation by NOT listening to our design efforts and modifying them accordingly.
How much NFB is good?
As a teenager in the late 60's, I built my first audio amps from vacuum tubes, and had my introduction to formal electronics by way of the Radio Amateur Handbook and an excellent book from and Argentinian engineer.
Vacuum tubes of course, where I learned about the benefits of negative feedback, though applied in modest amounts (less make and untamable oscilator!!).
Came the early 70's and unpaid work as student in the Electrical Engineering Department at the University, but the opportunity to work alongside heavyweights (Mr. Millman of textbook fame visited here among others).
Buzzword then was the operational amplifier concept, almost infinite gain and everything fixed by strong negative feedback.
In retrospect, and in the light of the current discussion from which I am learning lots for sure, one thing seems clear to me when dealing with what may be ordinarily considered second order issues but vital for high performance audio.
No matter how much NFB applied, nothing will really cure a gain path grossly nonlinear.
Seems to me much like trying in the real world to build a filter with perfect cancellation of certain pole with an appropiate zero. Will work in simulations but will degrade in production.
In the end, designers should strive for the best possible open loop linearity, and apply as much NFB as desired to improve further, not to cure an intrinsically nonlinear gain stage.
Rodolfo
As a teenager in the late 60's, I built my first audio amps from vacuum tubes, and had my introduction to formal electronics by way of the Radio Amateur Handbook and an excellent book from and Argentinian engineer.
Vacuum tubes of course, where I learned about the benefits of negative feedback, though applied in modest amounts (less make and untamable oscilator!!).
Came the early 70's and unpaid work as student in the Electrical Engineering Department at the University, but the opportunity to work alongside heavyweights (Mr. Millman of textbook fame visited here among others).
Buzzword then was the operational amplifier concept, almost infinite gain and everything fixed by strong negative feedback.
In retrospect, and in the light of the current discussion from which I am learning lots for sure, one thing seems clear to me when dealing with what may be ordinarily considered second order issues but vital for high performance audio.
No matter how much NFB applied, nothing will really cure a gain path grossly nonlinear.
Seems to me much like trying in the real world to build a filter with perfect cancellation of certain pole with an appropiate zero. Will work in simulations but will degrade in production.
In the end, designers should strive for the best possible open loop linearity, and apply as much NFB as desired to improve further, not to cure an intrinsically nonlinear gain stage.
Rodolfo
Absolutely right. Observe, theory, redesign, observe.
Don't be too sure about that.
True. But neither will an open-loop design. They use different applications of feedback and gain degeneration to compensate for the inherent non-linearities of the gain devices. I see no "principle" that judges the legitimacy of either approach. In my experience it is easier to get a good result by the brute force method (open-loop) but you can only go so far because of limitations introduced by this approach. NFB does not impose these limitations but demands much more precision of design and is more sensitive to imprecision; thus great designs are few and far between.
"What's your reasoning?" and not "What's your belief?"
My belief is the result, and not the reasoning.
My belief is the result, and not the reasoning.
John Curl,
We are not talking war politics here but rather distortions in audio amplifiers!
So why not try to keep things civil?
Clearly this discussion isn't going anywhere besides throwing sand at each other. While I may have the youth power you seem to have the experience so it would probably be an even fight! 😉
Now, I stumbled upon this article (Comparison of Nonlinear Distortion Measurement Methods by Audio Precision) which might interest people. http://www.audioprecision.com/index.php?page=resources&id=1000000162
You have to register to download it though.
While it does not address PIM it is quite interesting to see how well the good old THD method holds up (for TIM for example) if used properly.
Use it as you may see fit.
/Magnus
PS. SY, I love your avatar picture. Makes me smile everytime I see it. And that is with Chef's standards! 😀 DS.
Probably the first thing we will agree on. But I was not talking about history but the simple fact that I don't care about your personal issues expressed for example as the pointless insult "This is typical of academic types" in post #357. Some people may think that the name-calling of Curl, Otala, Gilbert et al along with detailed description of their sexlifes and a list of their favourite colors is an integral part of the history of electrical engineering - and in that case, good for them. I don't.
We are not talking war politics here but rather distortions in audio amplifiers!
So why not try to keep things civil? Clearly this discussion isn't going anywhere besides throwing sand at each other. While I may have the youth power you seem to have the experience so it would probably be an even fight! 😉
Now, I stumbled upon this article (Comparison of Nonlinear Distortion Measurement Methods by Audio Precision) which might interest people. http://www.audioprecision.com/index.php?page=resources&id=1000000162
You have to register to download it though.
While it does not address PIM it is quite interesting to see how well the good old THD method holds up (for TIM for example) if used properly.
Use it as you may see fit.
/Magnus
PS. SY, I love your avatar picture. Makes me smile everytime I see it. And that is with Chef's standards! 😀 DS.
- Status
- Not open for further replies.