odwello, others have been pointed to Cabot, Dunn, Hofer's AP manuals, whitepapers, refereed articles - and still want to do the strawman stupid audio engineer thing
I've also mentioned you can read Psychoacoustics - recent decades have seen many thousands of man hours of controlled ABX listening poured into perceptual lossy compression development
and while its a stretch it is practical to make power amps, DACs with errors below human hearing threshold in quiet limits
just a little use of critical band theory, human perceptual masking, reasonable numbers for home, recording venue, microphone noise floors let you be very confident that some distortions are simply belong human hearing abilities while music is playing at levels in quite mundane electronics'
I've also mentioned you can read Psychoacoustics - recent decades have seen many thousands of man hours of controlled ABX listening poured into perceptual lossy compression development
and while its a stretch it is practical to make power amps, DACs with errors below human hearing threshold in quiet limits
just a little use of critical band theory, human perceptual masking, reasonable numbers for home, recording venue, microphone noise floors let you be very confident that some distortions are simply belong human hearing abilities while music is playing at levels in quite mundane electronics'
Last edited:
Hiraga doesn't say, and I perceive from his graphics that he is not assigning
any numbers to this.
You are correct though, most of your more popular amplifiers historically
have had 2nd harmonic dominance at low power and segue to third
harmonic at higher power levels.
My own criteria is to look at the distortion at 1/10 rated power, and
regardless of absolute distortion level, I like to see about 10 dB less
3rd than 2nd, and I like 4th, 5th, and so on down in the dirt.
Like Pirate Rules, these be more by way of guidelines....
Which is the better?
1. Make open loop gain as high as possible then add large GNFB, or
2. Make open loop as linear as possible then add small GNFB.
1. Make open loop gain as high as possible then add large GNFB, or
2. Make open loop as linear as possible then add small GNFB.
Of course, Nelson, as you know, a remotely linear output circuit will always have MORE 3'rd harmonic at higher levels than at lower levels compared to the second harmonic. The second harmonic normally DOES dominate at extremely low levels, as this is the way an S curve generates distortion. IF you don't have that, then you have xover distortion, and that will be the dominant distortion of the amp.
It might be helpful in these discussions to remember hearing's sensitivity to non linear distortion decreases with increasing SPL. So that 1st Watt N. Pass fussed about is important - it's gotta be really clean.
Also, we need to remember that hearing's sensitivity to linear, or time domain distortions increases with increasing SPL.
You mention IM used by composers.
There is also AM. Humans can distinguish the beats caused by the cancellation/reinforcement of nearby tones up to about 20 times per second; beyond that, beats tend to fuse into whatever sound emitting process/object is creating them. This is Roughness and we are very sensitive to it up to rate of about 300Hz.
Roughness - fluctuation strength | Sound Quality | Acoustics Research Centre | School of Computing, Science & Engineering | University of Salford, Manchester
http://musicalgorithms.ewu.edu/learnmoresra/files/vassilakis2007smc.pdf
SRA 2.0 © 2008 - Spectral and Roughness Analysis of Sound Signals
Neat, eh?
.................................................................
Although at least half of musical energy is in the the area below 600 -700 Hz what we are really sensitive to is the range about 1- 5 kHz. So, I would want to know just how much of an amp's fm products are distributed into that 1-5kHz range because once they're there, they will sometimes modulate the vast numbers of musical overtones found there. This gets really sticky when reproducing large orchestral or choral forces.
That is a very interesting comment. I noticed that in my shop (where I primarily work on cars) that when I cranked the tunes with my old Nakamichi amp (receiver that I only use the amp), the mice would come out of every corner - but it sounded pretty good to me. My hearing is bad (not much after 10 kHz by my own measure) but other amplifiers, receivers, prototypes, etc) did not have the same effect. Many of them sounded better to my educated but badly degraded hearing, but ithe amplifier delivers the rated power and more (ratted 55 w/channel with 3 dB headroom per channel as measured by me with a pulse test). Maybe I didn't repair it correctly? Who knows; it's on the shelf again.
Last edited:
P
Good ol cats,
There could be an ounce of truth to this, or maybe not, we might never know, but funny enough my cat will place himself right between my Martin Logan ESL speakers driven by my Zero GNFb current feedback gain stage amplifiers of my own design :/. Yes they do use a meager 12db of loop feedback around the voltage gain stage, but this loop is isolated by an adaptive bias current stage. It is symmetrical and shares some jlh lineage, some depalma lineage and Diamond circuit lineage.
I pose this question to all, since we know dominant 2nd harmonic or. 3rd is acceptable(analog tape being rife with 3rd), what is an audible threshold of thd?. My thoughts have lay for a while that power amplifier thd is really the least important overall since it would be compounded immensely from the source chain, you would in reality need a preamp or source that measured perfectly. I love vinyl, but I measures terribly and with about 90-100db gain cartridge to speaker I don't think .1-1 % at the power end would even be slightly audible in this mess.
I view gnfb less as an absolute distortion reduction mechanism as I think this is musically better done by linearising before loop feedback, and using gnfb more to keep absolute closed loop again set and locked and maintaining offset control. Audio is fickle, it's an art, the stick by which we measure can be just as flawed as our hearing. As long as we follow voltage vs current laws there is no right or wrong.
Nelson, as far as 2nd dominating at lower levels I agree but would even say that close to equal 2nd and 3rd at low power levels with 3rd dominating at high levels would sound just as good.
Colin
Good ol cats,
There could be an ounce of truth to this, or maybe not, we might never know, but funny enough my cat will place himself right between my Martin Logan ESL speakers driven by my Zero GNFb current feedback gain stage amplifiers of my own design :/. Yes they do use a meager 12db of loop feedback around the voltage gain stage, but this loop is isolated by an adaptive bias current stage. It is symmetrical and shares some jlh lineage, some depalma lineage and Diamond circuit lineage.
I pose this question to all, since we know dominant 2nd harmonic or. 3rd is acceptable(analog tape being rife with 3rd), what is an audible threshold of thd?. My thoughts have lay for a while that power amplifier thd is really the least important overall since it would be compounded immensely from the source chain, you would in reality need a preamp or source that measured perfectly. I love vinyl, but I measures terribly and with about 90-100db gain cartridge to speaker I don't think .1-1 % at the power end would even be slightly audible in this mess.
I view gnfb less as an absolute distortion reduction mechanism as I think this is musically better done by linearising before loop feedback, and using gnfb more to keep absolute closed loop again set and locked and maintaining offset control. Audio is fickle, it's an art, the stick by which we measure can be just as flawed as our hearing. As long as we follow voltage vs current laws there is no right or wrong.
Nelson, as far as 2nd dominating at lower levels I agree but would even say that close to equal 2nd and 3rd at low power levels with 3rd dominating at high levels would sound just as good.
Colin
Last edited:
With a lot of shortcuts: he shows that the ear has a certain distortion mechanism that creates harmonic structure.
He then concludes that therefore, and amp must distort in the same way. I think the logical fallacy looms high over that.
Jan
That one sentence sums up the whole problem. Obviously this is an impossibility but even allowing for accidental choice of wording, how do you know that what you are measuring correlates with "not sounding different". There's been so much discussion already about how many decades ago a study concluded that THD is a very poor indicator of SQ. One smart-**** has insisted that "enlightened engineers" don't aim for lowest THD any more but has remained silent when asked what exactly they do use. [snip] Those that insist that listening to amplifiers is pointless and prefer to use test instruments to measure THD+N vs freq etc only to determine how good they are should stick to threads devoted to achieving best possible conventional measurements instead of trying to derail potentially interesting discussions with egotistical condescending statements implying anyone prepared to listen or approach amp design with an open mind is deluded or wants "tone controls".
I didn't mention the word 'measurement' so I am at a loss why you hitch on to that; I'll do you a favor and assume it is due to sloppy reading.
I believe the goal of 'sounding the same as the input signal' is exactly what the goal of HiFi should be, and that was my statement.
Jan
Last edited:
Jan I think we can agree that exactly that can not be evaluated with measument tools, then what do we have left apart from listening to the thing..??
True, but I think there still might be a possibility for a version of that theory in which you might want an amp (or a processor, or if I were a snarky person, an "effects box") to distort like the ear does in order to mask something else that is distorting in a more bothersome way. Such as (like I keep harping on) a loudspeaker that distorts significantly (at least compared to what can be easily done by an amp) in a very different, and frequency selective, way. A loudspeaker's distortion vs. frequency curve at higher levels (even of very good loudspeaker) can illustrate what I'm getting at.
Bill get your point, but if I would want to, for example, mask a distortion from a speaker, I would want an amp that is designed with harmonic products to specifically mask the spectrum from the speaker. I don't see why that should be the same as he distortion of the ear.
In fact, I DO NOT want an amp that somehow masks the expected distortion of the ear, as that would make the sound to differ from the original. The ear's distortion and performance is what it is and comes into play whether I listen to an original master or the reproduced version. So for audio designers, the ear's distortion does, to a first degree, not figure in it.
Jan
There's noting wrong with listening - in fact that would be the final arbiter wouldn't it? You can build the worlds best amp but if you and anybody else hates it, that's rather disappointing.
That said, I do think that we have the measurement armament to characterize an amp to check if it is HiFi for all intends and purposes.
I think about SY's test (Linear Audio Vol 2) of successively placing opamps in series and then listen until there is a difference in sound; IIRC his threshold was 5 or 6 in series.
We can do the same with power amps. Listen to a power amp, and then to a bunch of the same power amps in series. If, as an hypothesis, you cannot hear a difference between a single power amp or 5 or 6 in series, I'm willing to state that amp is HiFi in the original sense.
Jan
The distortion is a type that ears and brains are accustomed to, interpreted as being louder than without. The idea is use it to cover up speaker distortion, which ears and brains interpret as coming from a speaker (almost anyone can tell speakers apart by ear only, I believe).
Also, reproduced music is seldom played at original concert levels (inconvenient in homes/apartments, and most systems aren't capable).
I'm on about this after playing with some SE amps, and feeling like there is something to it. Yes its a distortion (easily seen even without diffmaker), but small, seen in scale of all else that is wrong in reproduced audio that can spoil the illusion. It could be a type of error that mitigates another.
BTW, your comment "The ear's distortion and performance is what it is and comes into play whether I listen to an original master or the reproduced version." reminds me of a similar argument I've heard (and even made) about reverb and reflections in a room. The recording has the reflections already there, the room shouldnt violate them. But have you ever heard reproduced audio in anechoic conditions?
Jan,
If we want hifi just build a blameless :/. Ears are the final judge, and you must admit there are very many variables. I have followed Diy Audio for 11 years now, something tells me no one here is going to be able to quantify the holy grail of audio based on the ideal of what we think as perfection. Some will buy gear based on the numbers, and others will fall in love with it's sound, I guess I'm the latter 😉.
Colin
If we want hifi just build a blameless :/. Ears are the final judge, and you must admit there are very many variables. I have followed Diy Audio for 11 years now, something tells me no one here is going to be able to quantify the holy grail of audio based on the ideal of what we think as perfection. Some will buy gear based on the numbers, and others will fall in love with it's sound, I guess I'm the latter 😉.
Colin
I understand that, I understand it's all about personal preference, and to a certain extend, anything goes. Somebody likes a BMW, anther swears by Dacia, I drive an Audi. I would never buy a Mercedes-Benz - why? I just don't like them.
If I would act in this area as a lot of people in audio, I would now go out of my way to convince people that M-B sucks, worst car ever made, and if you don't agree you are not just deaf but also blind and ignorant of cars.
Jan