As determined how? Measured thresholds of audibility? If so there has been considerable confusion among some engineers in regards to the correctness of their unstated assumptions.
You could use your honored listening capabilities and run a DBT test yourself to find out where your limits truely are (if they even exist).
This misses the point that many instruments have their timbre tweaked with distortion, but the mix between these separate sources is highly linear, decent mixing consoles have excellent linearity specifications. This means no intermodulation between guitar and vocals, for instance, even if each has been through separate effects - this means non-harmonically related intermodulation products are not generated. The human auditory system is very good at distinguishing separate sources from one another, but such intermodulation eats away at that ability and is undesirable. You cannot have second harmonic added without also creating intermodulation, and proponents of "even order harmonics = good" seem not to care about intermodulation between signals for some reason. Intermodulation products in most situations are not harmonically related to any of the contributing signals - they are dissonant, and typically numerous.
For instance try mixing vocals with raw guitar output, then passing through an overdrive pedal. The result is horrible. Pass them through separate pedals before mixing and you both have more control (separate control of each source), and intermodulation products between the sources are not created. Yes, in some contexts the artist might want that particular sort of horrible, but not generally!
If you are creating an amplifier for general use, it should be able to cleanly reproduce the source material, otherwise its not general purpose, its an effects unit that happens to also amplify. If you want that, fine, but don't call an effects unit a general purpose amplifier, such amplifiers have one job, to amplify, make bigger, louder. Adding such an amp to a signal chain shouldn't be detectable in a blind listening test - maybe adding 10 of them in a chain might become detectable, but one alone should be decently below the threshold of detectability to the human auditory system to be clean. Its like with windows, an optically flat clean window should be transparent, even stacking ten should be close to perfectly transparent. If it was slightly tinted that's detectable, and ten in a row would be awful! You might like rose-tinted windows, but that's a specialized visual effect, not a generic window.
After mixing comes mastering. Unless stems are used, all content gets the same treatment. Overall compression to taste and EQ are common final tweaks. Compressors do introduce dynamic distortions, with some units producing more coloration effects than others.
Beyond that, there is no ideal amplifier, not even Benchmark AHB2. I know because I have one. Its very good in some ways of course, but not the amp that gets used the most.
Beyond that, there is no ideal amplifier, not even Benchmark AHB2. I know because I have one. Its very good in some ways of course, but not the amp that gets used the most.
The question in the title was very simple: how do we perceive non linear distortions ?
Answer: if it's audible we perceive it as unwanted noise.
Harmonic distortions aren't non linear distortions so the topic should not spread over a thousand pages talking about THD its sources, audio mastering philosophy and so on with this title so my first suggestion to the thread suthor would be to change the title...We probably have a thousand topics and a few hundred thousands opinions on it...
If we want to talk about the "theory of everything" we should from the start make very clear difference between amp's distortions and speaker distortions, that's before aiming at thd imd crossover nonlinear distortions etc...
I would state first that distortions in genersl must be looked at also as from a reproduction system efficiency point of view exactly how an electrician sees the mains grid distortions.Why is that? Because low efficiency means less fundamental going through the system.
1% thd in 100watts system playing on 86db speakers is not the same thing as 1% thd in a 1watt system playing on 92db speakers. From here the maths guys can walk their way .I think there's an ex US Navy guy who recently published a study on the exact causes of each harmonic of THD...claming that's something we should know...probably we should know...i'm not sure of that though.
Some recent study saw that speaker's own nonlinear distortions tend to dissipate the amplifiers's THD and that is natural because energy is finite and in one cicle you only have some percentage of it available for destruction.The amplifier may impose its DF over the speaker to reproduce whatever the amp is telling the speaker to do or move, but amp's own harmonic or nonlinear distortions are not a wanted product by the amplifier , they are the result of poor amplification and poor feedback thus the amplifier's damping factor cannot act over its own THD, IMD or whatever other distortions might arise.
It's only the speaker who can act upon amp's THD.
The amplifier's crossover distortions imply that for a short period its damping factor goes either to zero or to a negative value so the speaker is left alone to damp itself while it will reproduce just some weird noises for a very short period of time and that might simply mean mechanically dissipating smplifier's distortions the way the dpeaker can do it so the rusult of it is not the amplifier's characteristic basically.
If the speaker is well damped by itself the duration of its own free oscillation will be as short as possible yet its back emf can affect the crossover period through feedback so the result is that in the crossover period we only hear the speaker damping itself in a more linear or nonlinear fashion depending on its design while the amp needs to get back on track, but crossover , if audible, will always sound like a noise echo or delayed noise.The shorter it is the better....With direct coupling, crossover can mean also DC imposed to speaker's coil so it will not be allowed to freely damp itself in a harmonic fashion anyway.
If we have tough loads we preffer only two species of amplifiers: ones capable of high amounts of audible low order THD to cover up for the rest of unwanted distortions (at least due to our own hearing time lenses capabilities which Thank God aren't too great and Ray Dolby explained us exactly why and how we can trick them even when playing exclusively the fundamentsl and tape's H3) and amplifiers with huge amounts of DF thus very tight feedback too . The best audio systems we know only exhibit inaufible distortions of any kind or audible H2 OR/AND H3, all the other higher order harmonic distortions being inaudible or covered up by h2-h3 .Our brain is asking for a narow harmonic image around fundamentsl in order to be able to keep on track otherwise our senses formed in millions of ears of evolutions get activated in order to hunt the noise source and we have a damn good computer for that.
Open baffle and horn speakers take advantage of a high and very predictible damping factor due to surrounding air thus dissipating the nonlinear or harmonic distortions in a very predictible and linear way while they exhibit fairly large efficiency /linearity ratio. We preffer them for tube amplifiers as tube amplifiers don't have great power or DF while their inherent low order high THD and zero dc output are allowing the speakers to dissipate the nonlinear distortions of any kind swiftly and in a very predictible way but that doesn't mean we want audible crossover at any sudio frequency in any system .We don't want zero DF .Unfortunately most of the measurements done on amplifiers are about how they behave at mid to high frequency, but i haven't seen many correlations between amp's DF at very low frequency and crossover distortions at high frequency...It msy be irrelevant as well too...
If we want more linear speakers of the modern kind with very low spl than the amplifier's specs needs to be grandious as they will need a fair amount of power, DF and slew rate to overcome speaker's requirements, but that may lead to either a very unpredictable oscillating system or a very neutral perfectly damped one that sounds borring as hell.
Not to mention motional feedback or sigma drive systems that keeps feedback so tight that you're only allowed to hear the recording medium and its decoding system and sometimes the recording itself is fairly bad, so bad that we need tubes or h3 tape distortions to listen to it...Motional feedback should be a more widespread solution if people would really like linear systems, but i'm afraid they don't really.I had a sigma drive system and it made some fairly low quality speakers sound like a million buck speaker system, but that tight control was so linear that i could hear the stupid music behind it and let's face it like men: we actually don't like a lot of the music we listen unless we do it on the "right setup "
...Now don't tell me that Neumann and Ortophone used motional feedback in their vinyl cutters because they wanted you to buy into more linear systems ...they simply wanted to keep the cutter inertia on the leash so that you can enjoy a vinyl...no matter how good or bad the music was on it.
Please keep in mind that i didn't choose to dive into why stereo systems are a fake image of what's on the stage that can be sold...If we really want the true sound on the stage we need as many speakers and tracks as the players are...thus we'd have a chance to less distortions, but sound engineers strive to put as many instruments on one channel while mixing on his 10 000 monitor speakers...which i bet most of you don't own so the argument of pure reproduction of the recording medium is completly busted from the start...The only guys who know how the recording was ment to sound are the artists and the mastering engineers, the rest of it is purely our imagination.
Answer: if it's audible we perceive it as unwanted noise.
Harmonic distortions aren't non linear distortions so the topic should not spread over a thousand pages talking about THD its sources, audio mastering philosophy and so on with this title so my first suggestion to the thread suthor would be to change the title...We probably have a thousand topics and a few hundred thousands opinions on it...
If we want to talk about the "theory of everything" we should from the start make very clear difference between amp's distortions and speaker distortions, that's before aiming at thd imd crossover nonlinear distortions etc...
I would state first that distortions in genersl must be looked at also as from a reproduction system efficiency point of view exactly how an electrician sees the mains grid distortions.Why is that? Because low efficiency means less fundamental going through the system.
1% thd in 100watts system playing on 86db speakers is not the same thing as 1% thd in a 1watt system playing on 92db speakers. From here the maths guys can walk their way .I think there's an ex US Navy guy who recently published a study on the exact causes of each harmonic of THD...claming that's something we should know...probably we should know...i'm not sure of that though.
Some recent study saw that speaker's own nonlinear distortions tend to dissipate the amplifiers's THD and that is natural because energy is finite and in one cicle you only have some percentage of it available for destruction.The amplifier may impose its DF over the speaker to reproduce whatever the amp is telling the speaker to do or move, but amp's own harmonic or nonlinear distortions are not a wanted product by the amplifier , they are the result of poor amplification and poor feedback thus the amplifier's damping factor cannot act over its own THD, IMD or whatever other distortions might arise.
It's only the speaker who can act upon amp's THD.
The amplifier's crossover distortions imply that for a short period its damping factor goes either to zero or to a negative value so the speaker is left alone to damp itself while it will reproduce just some weird noises for a very short period of time and that might simply mean mechanically dissipating smplifier's distortions the way the dpeaker can do it so the rusult of it is not the amplifier's characteristic basically.
If the speaker is well damped by itself the duration of its own free oscillation will be as short as possible yet its back emf can affect the crossover period through feedback so the result is that in the crossover period we only hear the speaker damping itself in a more linear or nonlinear fashion depending on its design while the amp needs to get back on track, but crossover , if audible, will always sound like a noise echo or delayed noise.The shorter it is the better....With direct coupling, crossover can mean also DC imposed to speaker's coil so it will not be allowed to freely damp itself in a harmonic fashion anyway.
If we have tough loads we preffer only two species of amplifiers: ones capable of high amounts of audible low order THD to cover up for the rest of unwanted distortions (at least due to our own hearing time lenses capabilities which Thank God aren't too great and Ray Dolby explained us exactly why and how we can trick them even when playing exclusively the fundamentsl and tape's H3) and amplifiers with huge amounts of DF thus very tight feedback too . The best audio systems we know only exhibit inaufible distortions of any kind or audible H2 OR/AND H3, all the other higher order harmonic distortions being inaudible or covered up by h2-h3 .Our brain is asking for a narow harmonic image around fundamentsl in order to be able to keep on track otherwise our senses formed in millions of ears of evolutions get activated in order to hunt the noise source and we have a damn good computer for that.
Open baffle and horn speakers take advantage of a high and very predictible damping factor due to surrounding air thus dissipating the nonlinear or harmonic distortions in a very predictible and linear way while they exhibit fairly large efficiency /linearity ratio. We preffer them for tube amplifiers as tube amplifiers don't have great power or DF while their inherent low order high THD and zero dc output are allowing the speakers to dissipate the nonlinear distortions of any kind swiftly and in a very predictible way but that doesn't mean we want audible crossover at any sudio frequency in any system .We don't want zero DF .Unfortunately most of the measurements done on amplifiers are about how they behave at mid to high frequency, but i haven't seen many correlations between amp's DF at very low frequency and crossover distortions at high frequency...It msy be irrelevant as well too...
If we want more linear speakers of the modern kind with very low spl than the amplifier's specs needs to be grandious as they will need a fair amount of power, DF and slew rate to overcome speaker's requirements, but that may lead to either a very unpredictable oscillating system or a very neutral perfectly damped one that sounds borring as hell.
Not to mention motional feedback or sigma drive systems that keeps feedback so tight that you're only allowed to hear the recording medium and its decoding system and sometimes the recording itself is fairly bad, so bad that we need tubes or h3 tape distortions to listen to it...Motional feedback should be a more widespread solution if people would really like linear systems, but i'm afraid they don't really.I had a sigma drive system and it made some fairly low quality speakers sound like a million buck speaker system, but that tight control was so linear that i could hear the stupid music behind it and let's face it like men: we actually don't like a lot of the music we listen unless we do it on the "right setup "
...Now don't tell me that Neumann and Ortophone used motional feedback in their vinyl cutters because they wanted you to buy into more linear systems ...they simply wanted to keep the cutter inertia on the leash so that you can enjoy a vinyl...no matter how good or bad the music was on it. Please keep in mind that i didn't choose to dive into why stereo systems are a fake image of what's on the stage that can be sold...If we really want the true sound on the stage we need as many speakers and tracks as the players are...thus we'd have a chance to less distortions, but sound engineers strive to put as many instruments on one channel while mixing on his 10 000 monitor speakers...which i bet most of you don't own so the argument of pure reproduction of the recording medium is completly busted from the start...The only guys who know how the recording was ment to sound are the artists and the mastering engineers, the rest of it is purely our imagination.
Mastering is a mixed bag for me... no pun.
I've got 24/88.2 files of great albums pre mastering that I much prefer to the mastered versions. OTOH really good ME's can just make the final result more polished and listenable.
Recently I had a fascinating discussion with a good friend that has been traveling SE Asia for many years recording tribal / native folk music direct to Nagra VII, stereo with some VHQ Schoeps mics.
His 2 track post 'mix' and 'mastering' skills have become so finely tuned, people have been sending him 2 track pop / indie work to re-master and fix.
It's really quite a black art manipulating a stereo mix to bring certain things forward, others back, widen some, focus others etc etc.
TCD
It seems to me that some members here only recognize their 'good' amplifiers, with low distortion, and 'effect boxes' with some form of distortion.
Lets say good amplifier is typical classAB amp 100 watts at 0.001% distortion.
Lets say effect box is typical classA amp 25 watts at 0.1% distortion.
One would assume outcome of listening session can be easily predicted. Low distortion amp will sound better. Well, not necassarily.
Low distortion amp has its lowest distortion before clipping, but its distortion progressively rises towards zero power.
ClassA on the other hand has its worst performance at max power, but the distortion is progressively going down towards zero at zero power.
In both cases we can not measure very low power as signal becomes noise dominant, but you get the point. ClassA amp towards zero will have better performance, while classAB will have increasing distortion towards zero.
I spent lots of time digitizing lp's and reel to reel tapes, using soundforge, signal crosses zero all the time. For 1 kHz, it crosses zero twice, two thousand times. Signal, just like membrane of the microphone, or a speaker, has to cross that zero all the time.
So those two amplifiers will have significantly different impact on the signal around zero. Sure at peaks classAB will be clean, but everytime signal approaches zero, it will distort much more than classA. And vice versa at peak power. ClassA will distort peaks, classAB will be clean.
Similarly intermodulation distortion, classA will aproach zero distortion every time signal aproaches zero. While classAB will generate significant intermodulation at low signals. Imaging suffers as a consequence.
Remember, no matter how quiet or loud you play, signal is crossing zero all the time.
So, at the end, there is no such thing as clean amplifier.
Scenario above has been done many times.
Only you can decide what you preffer.
Lets say good amplifier is typical classAB amp 100 watts at 0.001% distortion.
Lets say effect box is typical classA amp 25 watts at 0.1% distortion.
One would assume outcome of listening session can be easily predicted. Low distortion amp will sound better. Well, not necassarily.
Low distortion amp has its lowest distortion before clipping, but its distortion progressively rises towards zero power.
ClassA on the other hand has its worst performance at max power, but the distortion is progressively going down towards zero at zero power.
In both cases we can not measure very low power as signal becomes noise dominant, but you get the point. ClassA amp towards zero will have better performance, while classAB will have increasing distortion towards zero.
I spent lots of time digitizing lp's and reel to reel tapes, using soundforge, signal crosses zero all the time. For 1 kHz, it crosses zero twice, two thousand times. Signal, just like membrane of the microphone, or a speaker, has to cross that zero all the time.
So those two amplifiers will have significantly different impact on the signal around zero. Sure at peaks classAB will be clean, but everytime signal approaches zero, it will distort much more than classA. And vice versa at peak power. ClassA will distort peaks, classAB will be clean.
Similarly intermodulation distortion, classA will aproach zero distortion every time signal aproaches zero. While classAB will generate significant intermodulation at low signals. Imaging suffers as a consequence.
Remember, no matter how quiet or loud you play, signal is crossing zero all the time.
So, at the end, there is no such thing as clean amplifier.
Scenario above has been done many times.
Only you can decide what you preffer.
@adason Class AB means class A at low power...usually 1/10th of the nominal power....so a class AB will have zero crossover with soft signals but will also have twice the uncorrelated noise cause there are two devices in need for signal at higher bias and overal SNR is lower...with increasing component count in class ab you also need lower noise and tightly matched components.I remember someone telling me a decade ago about some noisy Krell amps...what we might actually enjoy more in between class a and class AB would be a very well matched class B amp, but that is really hard to get done with discrete components so class D took the challenge further.
Class D has introduced new problems that are slowly being better understood. For one example: https://purifi-audio.com/2020/04/28...n't something,but not entirely random moments. Lars Risbo said in an interview he doesn't think they have found everything yet.
Regarding Class AB, John Curl once biased an amp at 1A trying to get a large Class A region. Other designers have found perceptually optimum bias points that are amplifier specific. Takes ESL speakers to hear some of the subtle details more clearly.
Regarding Class AB, John Curl once biased an amp at 1A trying to get a large Class A region. Other designers have found perceptually optimum bias points that are amplifier specific. Takes ESL speakers to hear some of the subtle details more clearly.
I did something similar with a Realistic STA-2250 that i stll have, put a fan over the heatsink and indeed the sound became more pleasant, really pleasant.Unfortunately a passive cooling would have been really challenging if i'd wanted to have a listen at lower volumes without hearing the fan...That amp is open in the shed since 2011...never took the time to put it back together although its FM tuner is one of the best ever being also the second digital tuner ever made after STA 2100 made by the same company.Also STA2100 has one more filter.I should probably drag it again to class A cause its power transformer is 480watts while the power amps are 2x 50 watts class ab.The power supply is 2x 42v dc though...Not sure how much time will it survive in class A at 50 watts.The real chalenge though is the heatsink placement and its type... I have some server cooling fans with refrigerant liquids but i don't have the space for them inside the Realistic amp.The best amps of the 80's used freon for cooling which allowed for better cooling without active recirculating fluids.I still dream of building the ICE-AGE class A amplifier actively cooled with refrigerants...most probably someone else will do it before me , but talking about ice age a class B amp would probably be more appropriate to match.I' m pretty sure though that class A at room temperature would be a really interesting experience as tighter specs could be obtained.
Any nice distortion diagrams of push-pull tube amplifiers... class AB in particular?
Sounding better has many different definitions. It needs to be clarified with specific context. In the context of sound reproduction, it's the level of fidelity between input and output. In the context of personal preference, it's whatever the listener prefers. The latter creates problems when people are discussing it remotely such as internet forums. Why? I'm sure you've heard a phrase, "One man's trash is another man's treasure."
Well you just showed the class d purify module based nad c298 having the same distortions at low power as your class A gryphon...while the gryphon distortions rise with power same as in any class ab amp...not sure i'm getting your point anymore.