I assume many know how really harsh distortion sounds.
Very weak distortion may not be perceived the same way (i.e. harshness ar timbral change) but the IMD caused through the nonlinearity that is also responsible for the THD may rob the reproduction of clarity any "airyness".
Regards
Charles
Very weak distortion may not be perceived the same way (i.e. harshness ar timbral change) but the IMD caused through the nonlinearity that is also responsible for the THD may rob the reproduction of clarity any "airyness".
Regards
Charles
It's fairly clear that THD isn't a particularly good measure of much of anything. It certainly doesn't relate well to perceived quality or perceived "distortion."
There has been some recent work on percieved audibility of distortion which has shown that there is nil relationship between % THD and perceived audibility of distortion. There is a relationship, although it is a complex one (apparently) between the % of distortion per harmonic order and those order's relationships (ratios) and audibility - although a higher % of distortion does not mean higher audibility depending on which combination of harmonic orders are involved, in case you were wondering.
That researcher has shown some predictable results based upon the derived alogrithm of the research.
At least this is my understanding of it.
Having said that 0.001% THD is not a guarantee of inaudible artifacts at all. On the other hand 0.001% IM for a Cordell or similar test *might* be a better predictor, but I'm not going out on a limb with that one either.
The idea that putting an 80,000 ohm resistor in series with your speaker and that this corresponds to what you will hear as distortion % wise - isn't accurate or predictive of what actually happens. If it was then we'd still be not worried about jitter or >16bit, high clock rate digital reproduction.
The "wiggles" on a signal that is audible are not the same as trying to hear a signal at the threshold of audiblity (low in level).
Although, if you sit in a sufficiently quiet room, you'd be rather surprised at what you *can* hear once your hearing shifts...
... my 2 cents...
_-_-bear
There has been some recent work on percieved audibility of distortion which has shown that there is nil relationship between % THD and perceived audibility of distortion. There is a relationship, although it is a complex one (apparently) between the % of distortion per harmonic order and those order's relationships (ratios) and audibility - although a higher % of distortion does not mean higher audibility depending on which combination of harmonic orders are involved, in case you were wondering.
That researcher has shown some predictable results based upon the derived alogrithm of the research.
At least this is my understanding of it.
Having said that 0.001% THD is not a guarantee of inaudible artifacts at all. On the other hand 0.001% IM for a Cordell or similar test *might* be a better predictor, but I'm not going out on a limb with that one either.
The idea that putting an 80,000 ohm resistor in series with your speaker and that this corresponds to what you will hear as distortion % wise - isn't accurate or predictive of what actually happens. If it was then we'd still be not worried about jitter or >16bit, high clock rate digital reproduction.
The "wiggles" on a signal that is audible are not the same as trying to hear a signal at the threshold of audiblity (low in level).
Although, if you sit in a sufficiently quiet room, you'd be rather surprised at what you *can* hear once your hearing shifts...
... my 2 cents...
_-_-bear
As for most amps I've tested, there isn't much correlation between power and distortion. Generally if distortion is low at high power, it's low at low power as well.
I know a certain lurker on here, who shall remain nameless, who claims that amplifier that may have low distortion at high power but have much high distortion at lower power. He's trying to capitalize on the audiofools who will believe it.
It's kind of hard to swing from a high negative voltage to a high positive one without going through the distortion prone crossover point. It's more common to see an increase in distortion with power, particularly with class A amplifiers. All semiconductos that I'm aware of lose linearity at high current, rather than low current. You'll never hear somebody talk about distortion caused by transistors 'sticking' in cut-off.
THD+N and IMD numbers work perfectly well for describing the quality of an amplifier. Channel seperation is an often overlooked one, especially with digital sources. Numbers < 0.1% for THD+N and < 0.05% for IMD seem to be the rule. If you can stay < 0.05% for both, from say 0.25 watts and up, you've got a really good amp.
I know a certain lurker on here, who shall remain nameless, who claims that amplifier that may have low distortion at high power but have much high distortion at lower power. He's trying to capitalize on the audiofools who will believe it.
It's kind of hard to swing from a high negative voltage to a high positive one without going through the distortion prone crossover point. It's more common to see an increase in distortion with power, particularly with class A amplifiers. All semiconductos that I'm aware of lose linearity at high current, rather than low current. You'll never hear somebody talk about distortion caused by transistors 'sticking' in cut-off.
THD+N and IMD numbers work perfectly well for describing the quality of an amplifier. Channel seperation is an often overlooked one, especially with digital sources. Numbers < 0.1% for THD+N and < 0.05% for IMD seem to be the rule. If you can stay < 0.05% for both, from say 0.25 watts and up, you've got a really good amp.
Folks, the facts are that it is easy to make an amp that measures very well at high levels, and poorly at low levels. Just make it class B, which usually generates some crossover distortion. High feedback will tend to hide distortion, but crossover distortion, if serious enough, implies a dead zone, that actually removes gain from the forward path at crossover. This makes things worse than normally expected.
Also, crossover distortion is often hidden in the residual noise (80KHz or so) of the test equipment measurement at low levels, and of course, is reduced in relative amplitude when measuring high signal levels, becaise the crossover distortion region remains the same level, and the ratio between the two is greatly increased, compared to low signal levels
Also, crossover distortion is often hidden in the residual noise (80KHz or so) of the test equipment measurement at low levels, and of course, is reduced in relative amplitude when measuring high signal levels, becaise the crossover distortion region remains the same level, and the ratio between the two is greatly increased, compared to low signal levels
Without getting in to bickering what is behind some of this i:
A- for a fe volts below vsible clipping on an oscilloscope, distortion starts to rise i beieve nearly everthing behaves his way however.
and
B- at really low levels the noise loor becomes a issue. Nothing really out of the ordnary here. I appear wose wih ver high power/gain amps if only because i originates in the nput tage ad then gets amplifid to signiicant levels.
Sometime thi gets ino an EF vs. CFB debate sinsecrossover distortion is saidt increase or decrease oppositly suith each opology -- I can't remember which is which and don't realy care becase there are more significant hings to consider incoosing bewen the two.
Only in some regards. I once read an article about the hearing capabilities of owls. Namely, their ability to locate prey by "stereoscopic hearing" at night. The article said that while human bandwidth was limited, within the bandwidth, the only animals that did better than us were owls. Others may hear at higher or lower frequency or at lower levels, but passive location (NOT echo location like bats) we come in number two after owls. Perhaps this correlates with the obsession for "sound stage" amoung audiophiles.
maylar said:
Given that there is no consensus about this in "audio circles", there is a simple answer: "no".
However, to say that THD measurements, or measurements in general, are disconnected from audible differences between various audio components is nonsense. Regrettably, I think that many a debate has been muddled by the confusion of "audible differences" and "objectionable differences" or "objectionable impact". Critical limits of audibility for THD have of course been established but strictly speaking, they are only valid under the precise set circumstances from which they were derived, such as program material, spectrum, and other nonlinearities in the system. Thus, they vary somewhat. However, to use such information as a universally applicable limit to judge the audible "signature" of an amplifier is not advisable from either a scientific or a practical point of view. But given that the level of THD is pushed under a certain level that is well below the lowest threshold that has been demonstrated in scientific experiments, it is reasonable to assume that the THD level would not the first place to look for explanations to audible differences. There are so many other things affecting the sound that one measurement alone cannot possibly capture it all.
What is said above is a general observation. Individual differences in the sensibility to various nonlinearities will play a decisive role in specific cases. Casual observation of some reviews in The Stereophile is enough to demonstrate that people differ vastly in their sensibility to distortion (or in their ability to overlook it).
Sincerely,
Kranis
john curl said:
I take it you mean that negative feedback "reduce", not "hide" distortion in the complete amplifier.
john curl said:
How do you know it is there if it is below the noise floor of your test equipment?
But seriously, if the crossover distortion produced from an optimally biased class B amp is pushed below the noise floor, the same effect would take place in a class A amp: the S/N ratio will decrease at really low levels as noise is constant but the signal decreases. THD would fall monotonically in both cases as the output level is reduced (but would be lower for the class A amp) until THD reaches the noise floor. Or did I miss something here?/Kranis
Mood meter together satisfaction Meter
When sound is distorted, at least 0.6 percent...you do not feel confortable....you will have two good moments...the spectation when you switch it on, and the happyness when you switch it off.
It turns you nervous.... you cannot think in anything, cannot talk...nothing...the damn thing dominate you entirelly, the brain have enormous work to processing all that mass...and transform in something agreable to your conscious...conscience....English!... aaagh...complicated.... felling good.... knowing the sound is good...having knowledge the sound is good.
We do not know almost nothing about Brain...so..there are room enougth to all hipothesis and theories.
When the sound is too much bad, as one terrible Technichs hi top level, incredible surround, wonderfull panel... ten billion transistors.... 200 chips..... and more 50 transistor for audio and 2 chips to develop power.
I made three speakers different....i bougth two more...and always changing the equalization...never satisfied....never good enougth, and the one, we could adjust each frequency....not increasing 1 Kilohertz and 500 hertz beeing increased together..
It turned myself crazy...i sold it .... a happy day...seeing that "thing" going away.
This is distortion, bad sound...we react adjusting and never satisfied.... and we are happy to switch off...if you perceive the sound playing in low volumes.....always it appearing...it may be bad...as disturbing, not making you dream...not relaxing you.... disturbing you.
regards,
Carlos
When sound is distorted, at least 0.6 percent...you do not feel confortable....you will have two good moments...the spectation when you switch it on, and the happyness when you switch it off.
It turns you nervous.... you cannot think in anything, cannot talk...nothing...the damn thing dominate you entirelly, the brain have enormous work to processing all that mass...and transform in something agreable to your conscious...conscience....English!... aaagh...complicated.... felling good.... knowing the sound is good...having knowledge the sound is good.
We do not know almost nothing about Brain...so..there are room enougth to all hipothesis and theories.
When the sound is too much bad, as one terrible Technichs hi top level, incredible surround, wonderfull panel... ten billion transistors.... 200 chips..... and more 50 transistor for audio and 2 chips to develop power.
I made three speakers different....i bougth two more...and always changing the equalization...never satisfied....never good enougth, and the one, we could adjust each frequency....not increasing 1 Kilohertz and 500 hertz beeing increased together..
It turned myself crazy...i sold it .... a happy day...seeing that "thing" going away.
This is distortion, bad sound...we react adjusting and never satisfied.... and we are happy to switch off...if you perceive the sound playing in low volumes.....always it appearing...it may be bad...as disturbing, not making you dream...not relaxing you.... disturbing you.
regards,
Carlos
Kranis, have you made any THD measurements? In order to resolve low level distortion created by crossover distortion artifacts, we usually use some form of spectral analysis on the THD+N residual and reduce the noise floor, perhaps 20-40dB.
Generally, crossover distortion is separate from class B distortion, and rises proportionally at low levels. It is not monotonic with level.
Generally, crossover distortion is separate from class B distortion, and rises proportionally at low levels. It is not monotonic with level.
Nelson Pass said:The other hallmark of a good system is not having to crank it up.
True, but when old Karl Richter stepped over the organ's pedals, I like to feel it in the chest. Even enjoy some mild rattling of things arround as long as it does not come from the speakers themselves.
Rodolfo
Rodolpho, please inform what is crank
Can you please inform to a poor brazilian, completelly lost with those different words...my dictionary is smelling, fungus, because i live in front of sea...and paper turn bad smell.
O que és Crank, usted puede explicar para tu hermano sud'americano, por favor.
Carlos
Can you please inform to a poor brazilian, completelly lost with those different words...my dictionary is smelling, fungus, because i live in front of sea...and paper turn bad smell.
O que és Crank, usted puede explicar para tu hermano sud'americano, por favor.
Carlos
Please write to my adress, to not disturb thread with my own problem
My personal E mail is:
nanabrother@yahoo.com
And you will welcome
regards,
Carlos
My personal E mail is:
nanabrother@yahoo.com
And you will welcome
regards,
Carlos
Hi Bear,
You said it, but no one responded - " The wiggles in the sound - "
Why is everyone dazzled by ultra low and resistor measured THD and IMD figures observed in time delayed sinusiodal isolation ?
Those *phase shifted* (i.e. time shifted wrt the original waveform that generates them) wiggles are caused by *music* driven dynamic crossover network/driver back EMF interacting with output stage conduction crossover + the phase/level of NFB loop controlled damping + amplifier (stabilisation) propagation delay.
These waveform modulating wiggles are less easily masked because they are additional, not harmonic, and they still arise due to propagation delayed NFB response allowing a non-class-A output stage to be suddenly reverse commutated through a portion of its class-B bias voltage, even after the forward measured sinusoidal 'crossover distortion' has been pushed below the noise floor !!!
If anyone does not believe me, then listen to an amplifier with headphones from another room as the output loading is switched between a resistor and a real world loudspeaker. Resistor operation will be as the THD specification suggests; but the THD specification will not be similarly relevant to loudspeaker loaded operation.
This is why a non-feedback amplifier with quality full-range driver can be so much less aurally disturbing.
Cheers ........... Graham.
You said it, but no one responded - " The wiggles in the sound - "
Why is everyone dazzled by ultra low and resistor measured THD and IMD figures observed in time delayed sinusiodal isolation ?
Those *phase shifted* (i.e. time shifted wrt the original waveform that generates them) wiggles are caused by *music* driven dynamic crossover network/driver back EMF interacting with output stage conduction crossover + the phase/level of NFB loop controlled damping + amplifier (stabilisation) propagation delay.
These waveform modulating wiggles are less easily masked because they are additional, not harmonic, and they still arise due to propagation delayed NFB response allowing a non-class-A output stage to be suddenly reverse commutated through a portion of its class-B bias voltage, even after the forward measured sinusoidal 'crossover distortion' has been pushed below the noise floor !!!
If anyone does not believe me, then listen to an amplifier with headphones from another room as the output loading is switched between a resistor and a real world loudspeaker. Resistor operation will be as the THD specification suggests; but the THD specification will not be similarly relevant to loudspeaker loaded operation.
This is why a non-feedback amplifier with quality full-range driver can be so much less aurally disturbing.
Cheers ........... Graham.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.