Find us a polar electrolytic that is accurately modeled only with ideal R and C.
-RNM
Relevance to what I said? NPO ceramic has .6% DA and virtually no distortion, see Samuel Groner's measurements. Or see Bob Peases' article on DA. Distortion and DA are not closely linked and presenting pathological cases does not prove a point. Just like using 1/4W resistors in a 400W PA feedback network, it's just bad design not the resistor's fault.
I repeat again all measured results for precision amplifiers, DAC's, and A/D's are taken and verified with an AP or equivalent and are done with ordinary off the shelf parts.
Last edited:
What does the term "hidden variables/information" really means in this context? And what about the "etc."? If it was inserted to cover claims being contradictionary to first principles and cases where really overwhelming experimental evidence exists, i can agree otherwise it needs clarification too. (stubbornly i know)
Sensu stricto, "hidden variables" theories come from quantum physics, as an attempt to avoid the indeterminism in predicting the outcome of a measurement.
The famous Bell theorem showed that if local hidden variables would exist, then there are some quantum entanglement experiments that would obey the Bell inequalities. However, statistical correlations resulting from quantum entanglement would violate the Bell inequalities, so they cannot be explained by local hidden variables. As a corollary, the existence of local hidden variables is equivalent with accepting superluminal interactions.
By extension, "hidden variables" means (to me) invoking all kind of root causes for audible differences, some of which don't make any sense by any metric of the current body of knowledge. You have an example above with the "dielectric" being responsible for audible differences between mains cables. From the same author, for whatever reason ("hidden variable") bitwise identical digital files can sound different, depending on the support they are stored on (e.g. thumb drive vs. hard drive).
Perhaps the mains cable example is not the best for exemplifying the outrageous claims made by the Golden Ears, but I'm sure you got the drift.
Many claims of audio product have been debunked via DBT or measurements. As the saying goes, cry wolf once...
Perhaps we could say nonlinear-distortion and DA are not closely linked? Not clear if anyone has determined that linear distortion of a ladder network cannot be audible.
I only make the point because part of the dispute about what people may think they hear and any associated physical processes in a some capacitors could be affected by how the term distortion is being defined by everyone in the conversation.
Last edited:
I challenge the concept of “linear distortion”, in particular in audio, but that’s me, no need to start a new religious war over this.
Doesn’t matter how you slice it or dice it, audio amplifiers are minimum phase systems, at least until Ft or other residual poles frequencies are reached (10s - 100s MHz). As such, “linear distortions” are nothing but bandwidth limitations (and the associated phase).
Same story with the ladder network; if you want to call its LPF effect as “linear distortion” that’s your call, but the audibility has nothing to do with the LPF concept (which always has “linear distortion”), but with the LPF frequency response (and phase).
“Linear distortion” is IMO a derivate and useless concept even in communications, where the phase response of the system can be critical.
So, if i am not mistaken, it is a bit of redundancy at this point in the list, right?
Nothing wrong with that as long as you don´t take it for granted....
Glad that you posted a remark about the inherent limitations. I should have posted it together with a disclaimer but initially thought the material would naturally be taken in the context of the year 1988 (and devices from that time), but of course that is/was not ensured.
Not my term, its an old idea which can be found in some textbooks. The idea is any unwanted alteration to a signal is either noise or distortion, and distortion can be separated into linear and non-linear types.
So, if i am not mistaken, it is a bit of redundancy at this point in the list, right?
Nothing wrong with that as long as you don´t take it for granted....
Glad that you posted a remark about the inherent limitations. I should have posted it together with a disclaimer but initially thought the material would naturally be taken in the context of the year 1988 (and devices from that time), but of course that is/was not ensured.
Fair enough.
Of course.
It remains a guide line, the details of making the sound clips are given in most cases, possibly someone could recreate them with SOTA equipment. A couple of us here made a recording of a solo performer (voice, harmonica, guitar. 24/96) as an experiment with no limiters or equalization of any kind (mics directly into A/D). The clarity of the harmonica was especially good to my ears, but that is through the <$300 field recorder and headphones.
If people "miss" them because they can't hear a difference, than they are inaudible, by definition.
In this case again, when examined the phase and envelope "distortions" were fractions of a degree and .01 or less dB in most cases. Studies of audibility of phase sometime use all pass networks with gross phase deviations, not the same thing.
The last time we discussed DA the conversation degenerated into semantic arguments over things like the meaning of perfect vs ideal components.
My goal is certainly not to create more arguments. It is just that there have been some arguments in the past about something not sounding right, or maybe sounding time-smeared (whatever that means, not sure), and other people retorting that it must be impossible because there is no measurable distortion when in fact they are referring to unmeasurable nonlinear distortion.
Can a ladder network ever sound like something that may be perceptually referred to as 'time-smeared?' I don't know, but don't know that it can't either. My preference, rather than always assuming people must be imagining something would be to defer judgment until we know more, even if that may not be in my lifetime.
But, for convenience sake I am happy with the shorthand that the term distortion usually refers to nonlinear distortion, so long it produces less arguments and less jumping to conclusions than other options.
I doubt there is such a thing these days (WRT audio), as I said earlier the problem lies in the reduction and interpretation of data we already know how to take. These issues get this thread off track, are we really stuck again on the issue of blind testing.
No, that's what I meant. I would agree that anything people can hear should show up in some kind of measurement so long as we happen to be looking for it. What I meant was that if no nonlinear distortion is found, then some people tend to conclude there can be nothing objectionable about sound quality. Sorry, but group delay sometimes sounds bad, depending, which is only one example. The problem there is the jumping to conclusions inappropriately. We still have people saying 2nd HD above 10kHz can't matter, but they ignore the IMD that can also occur. We still have people using THD@1KHz as a proxy for distortion in general. And, lot of arguments are based more on long ago formed conclusions than on the substance of what is being offered up for argument. Lots of people problems, probably more so than technical problems.
Last edited:
We should not forget that there is a distinction between (possible) audible differences and the alleged reasons for this (possibly existent) audible difference.
To follow the example, one can not rule out that on some machines there exists an audible difference when bitwise identical files are replayed/routed from different hardware devices, but one can rule out that the reason is some kind of a "ghost in the machine" .
Sometimes it is better to be more precise, if one really wants to sort it out.
Last edited:
Maybe, but minimum phase systems (as all audio amplifiers) have the minimum possible group delay. So good or bad sounding, you can't do a better amplifier, other than increasing the bandwidth.
Now, if you are talking about speakers, that's a different kettle of fish, since speakers are in general not minimum phase systems.
About distortions, I still have to find a competent engineer looking at THD @ 1KHz only. Those who rely on such metric need to go back to school.
Specifically about 2nd harmonic @ >10KHz, a simple Taylor expansion will show how quickly the intermodulation products amplitudes drop. So it's about the HF nonlinearity magnitude, and one that would create audible IMD products in the ear's most sensitive frequency range would need to be rather gross. See the multiple examples in the Atkinson/Stereophile much beloved tube amps measurements. Any competently designed current solid state amplifier will have intermodulation products that are way under any audibility threshold.
Some will nitpick about the "way under any audibility threshold" above, as an assumption. I would only point them to the High End Audio Priests which consider, for a boutique power amp, 0.01% THD 20Hz-20KHz as nothing less is, and will, ever needed. So why further worry about the (always lower) IMD components sizes?
Yes it can. You just stepped on a mine, explaining such differences would collide with the thermodynamics principles, violate Shannon, etc... One doesn't need to measure a difference, it's enough to assume such a measurement exists and physics, as we know it, explodes in his hands.
Otherwise said, if any sensory testing shows differences, then you'd better try to find what's wrong with the test, rather than assuming the impossible.
If you want to spend the rest of your life doing such stupid sensory tests, you are my guest. But don't expect anyone to take you seriously, other than a few that have a vested interest in selling FUD countermeasures to those with deep pockets and shallow gyri.
Please note that nobody knows what participants in test _really_ have heard, as we can´t assess it directly.
All we have are results from a listening experiment, so the usual analysis only tells us if these observed results are more or less compatible to the assumption that the test listeners might have guessed randomly.
But that does not tell you, if the observed results exists due to the conditions of the test or because _these_ listener really can´t (means never) detect the difference.
As stated before, we already know that the same test subjects get significantly _less_ correct trials when probing the same sensory difference in an ABX test than in an (for example) A/B test.
I hope that explains, why your above cited conclusion is a "non sequitur".
- Status
- Not open for further replies.