Pan said:
I'm sorry, I thought it was clear with the quotes. Dr. Geddes (the Dr distinction is important as PhD doesn't stand for "Pizza Hut Delivery") stated "High levels of low order nonlinear distortion are virtually always viewed as more pleasing." You subsequently gave an absolute "Nope" without providing any technical references or elaboration upon your claims outside of a derivation of "I hear it therefore it is."
The article I linked to is in nearly its entirety a technical discussion of harmonic distortion and perception. The basic premise being that low order nonlinear distortion is almost always viewed as more pleasing where as high order distortion is that which the ear is most sensitive to. The article even elaborates on a claim that higher levels of THD can sound better than lower levels if the higher order harmonic distortion is suppressed compared to lower orders. Though it does contain a few too many equations to be easily, casually read, I thought pretty informative.
In other words, it supports Dr. Geddes' argument and diminishes yours unless you have your own references to provide that are something more than "Nope."
- JP
Diogenes said:
Oh sorry.. but we Swedes do not pay much attention to titles and stuff.. just the way we are.
The article is pretty basic, no news.
This is what gedlee wrote;
I did read that first sentence in isolation.. maybe I didn't grasp the meaning?
I read it as high levels of low order harmonics is more pleasing than would they not be there.. period.
I do agree that high levels of low order harmonics are more "pleasing" than high levels of high order distortion. If that is what gedlee meant.
Now, regarding my "Nope". I have done blind listening tests with electronics that allows for adding distortion (by altering the gain in some stages) and my conclusions from those tests, among with other tests and other peoples tests, is that transparency is prefered before coloration.
The low distortion poweramps I have blindtested have always given a more pleasing and transparent result than the amps that gives 0.5-1% THD (dominantly 2nd and 3rd and at aprox. 1W).
Also want to mention that I realise that in some occasions a certain program material MAY sound subjectively better to some ears with some audible distortion thrown in.
My experience recording my own voice and steel strings point at transparency though.
/Peter
Diogenes said:
Your summary is that the article is "a technical discussion of harmonic distortion and perception." I think you might be confused in your summary (see portions highlighted in red).
But assuming your summary is correct, this has nothing to do with Earl's assertion, which is about perceptions of various levels of NON-LINEAR distortion. He does give one example that is non-linear (crossover distortion). Need I point out that harmonic distortion is a type of linear distortion, not a type of non-linear distortion?
I found myself puzzled by his statement anyways: since when is non-linear distortion ordered? Is he referring to the sum tones of IMD as orders? Something peculiar to his own "higher order modes"?
- Eric
PS - PhD also stands for "piled higher and deeper." My apologies to the good doctors amongst us.
Pan said:
Agreed if that was the assertion. Since there is no such thing as "distortion" in live music, or any other original (as opposed to reproduced/processed) sound, any audible distortion of any kind means you are not hearing the same original sound - you are hearing something different.
sdclc126 said:
Agreed. Learn to hear it, then identify it, then remove it, and you'll be home free. See my earlier post in this thread, where I tried to show the common ground between Lynn Olson's prescription and Earl's objections to listening to live classical music. Whatever happened to Lynn's participation in this thread, anyways?
- Eric
Pan said:
Perhaps common, but not supported.
I've looked far and wide found no formal definitions for linear and non-linear distortion. I find what you wrote in many places. The oldest source I have is RDH4, where Langsford-Smith provides a list of these types of distortion
1. Non-linear distortion (also know as amplitude distortion), and resulting in
1a. Harmonic distortion and
1b. Intermodulation distortion.
2. Frequency distortion (unequal amplification of all frequencies)
3. Phase distortion.
4. Transient distortion.
5. Scale distortion (or volume distortion).
6. Frequency modulation distortion.
He cites Olson a lot in the rest of the chapter the contains this list.
The only rationale I can find for the HD and IMD (1a and 1b in the list) being called non-linear is that they are produced by non-linear spacing of the amplifying device's characteristic curves at some particular operating point/load line. I've read one analysis that attributes HD to this, and another that attributes IMD to this.
So we have a potential reason for naming two distortion types, but (1) this reason only applies to amplifiers and (2) categorizing two types as non-linear doesn't confer linearity on the other types.
(BTW, I've looked many other places. Horowitz, Olson, etc. Interestingly, Olson does some handwaving in passing. In the Measurements chapter in Acoustical Engineering, he says, "The nonlinear distortion characteristic of a loudspeaker is a plot of the total distortion in per cent versus the frequency at a specified power." From nonlinear distortion to total distortion. None of his other sections on distortion shed any light on the nomenclature either.)
So where to look for an answer? Linear systems. For example, where the nonlinearity of air causes problems for compression drivers, it introduces harmonic distortion. So we need to look at the linear (or nonlinear) properties of each part of the system. Fortunately, the definition of a linear system is very simple, if hard to describe with great precision.
Given a complex input signal, a linear system will produce a response that is identical to the sum of the system's responses to each of the components of the complex input. Or, A(x1) + A(x2) = A(x1 + x2) where A() is the response of the system, x1 and x2 are the inputs, and "+" means the inputs are applied simultaneously.
If a speaker produces harmonic series for each of two separate pure tone inputs, and if given the same two tones simultaneously, it produces an output that is the sum of the two harmonic series, then the speaker is a linear system behaving linearly, even though it's generating harmonics.
Likewise, if a speaker generates the harmonic series for a pure tone, and when the level of the tone is doubled, the level of the harmonics are doubled, it's behaving linearly.
If new tones are created when the two tones are input, (intermodulation distortion) the speaker is not behaving as a linear system. I have read phrases like "harmonic distortion and nonlinear intermodulation distortion" several times in many places where the issue seems to be sidestepped.
Keep in mind that the actual transfer function of the linear system doesn't determine if it's linear or not. If the system always adds harmonics, it can still be linear. The production of harmonics by whatever mechanism says nothing about whether this distortion is linear or nonlinear. Think of a synthesizer.
Please let me know if you have any references that define nonlinear distortion beyond "addition of new tones". Heck, I can't even find one using google or wikipedia! I wonder if Olson or someone ever used dynamical analogies to map the physical behaviors that cause driver distortion to electrical models and found that the results looked a lot like some triode.
- Eric
Eric,
But if the transfer curve is such that new harmonics are produced on a single sine, then two sines will produce IM. Also it seems funny to call it "non linear IM" since as far as I understand there no such thing as "linear IM". Would the system be linear there would be no IM.
IM is also not only sum tones but also difference tones.
Interesting thoughts though and I'll have to digest some of what you wrote.
/Peter
But if the transfer curve is such that new harmonics are produced on a single sine, then two sines will produce IM. Also it seems funny to call it "non linear IM" since as far as I understand there no such thing as "linear IM". Would the system be linear there would be no IM.
IM is also not only sum tones but also difference tones.
Interesting thoughts though and I'll have to digest some of what you wrote.
/Peter
IMO the common definition of linear/nonlinear in EE implies that we talk about the DC or time domain transfer function. Distortion there is anything that is not linear, hence, "linear distortion" is not applicable. This term is used when another deviation from the ideal is examined and "linear" refers to a property that can be corrected later on (which is impossible with distorted DC/time transfer function unless you want to look at the signal upstream). And a linear system is yet another thing (just like Eric pointed out). A speaker, at small signals, can truly be a linear system, with a bit of goodwill even linear time-invariant (LTI), but still it can have non-linear transfer function and overall nonlinear frequency response (which usually only looks at the fundamental, per definition)
It's somewhat a semantics problem, in case of doubt "linear" should be qualified (referring to what exactly is linear or not) .
- Klaus
It's somewhat a semantics problem, in case of doubt "linear" should be qualified (referring to what exactly is linear or not) .
- Klaus
ShinOBIWAN said:
I've had a good listen to the Orion including unedited master tapes of classical recordings. I definitely couldn't say the bass was a revelation compared to other bass speakers I have heard. It was like any other good bass setup, which excludes all those systems with the bass overboosted.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.