Sometimes it really helps to actually read linked PDFs.
Hint - Jazz Man's post included a link to a publication from Rousseau/Ennis well worth to read, if one is really interested in good experiments. It includes a short but good overview about different theories and a nice reference list is included as well.
Hint - Jazz Man's post included a link to a publication from Rousseau/Ennis well worth to read, if one is really interested in good experiments. It includes a short but good overview about different theories and a nice reference list is included as well.
Fully agree, it´s very hard to prove wild speculations fully opposite to established knowledge which is used everyday, everywhere, by experienced and established Manufacturers.
Not impossible, but they have to start by having something real to prove.
Not impossible, but they have to start by having something real to prove.
In the case of resistors it probably comes down to something simple, an occam type of explanation. We already know resistors can have low level signal correlated excess noise. There is some evidence, not ideal evidence, but still some that says some humans can notice very low levels of signal correlated noise. Most likely between that and a bit of HD in some resistors, some people notice something about the sound of some resistors that is enough pass a double blind test.
It wouldn't be Earth shattering if that turned out to be what all the arguing has been about over the past decades. Some physical property we already know about, but that we have ASSUMED nobody could possibly hear.
The other popular explanation is that it is all imaginary and the double blind test results never happened.
Just for something else to think about, most of us have read in an audio forum about a supposed 'expectation bias' that means people hear what they expect to hear. While I am not convinced any such bias has been properly studied for hi fi sound reproduction, for those that choose to believe in such a bias please me suggest how it might work: Someone sees FFTs showing an audio device has noise and distortion down at -120dB or lower. Some people expect with measurements like that the SQ must be audibly perfect. Since those people expect perfection, that's what they hear, even if the sound isn't really all that perfect to an unbiased listener.
It wouldn't be Earth shattering if that turned out to be what all the arguing has been about over the past decades. Some physical property we already know about, but that we have ASSUMED nobody could possibly hear.
The other popular explanation is that it is all imaginary and the double blind test results never happened.
Just for something else to think about, most of us have read in an audio forum about a supposed 'expectation bias' that means people hear what they expect to hear. While I am not convinced any such bias has been properly studied for hi fi sound reproduction, for those that choose to believe in such a bias please me suggest how it might work: Someone sees FFTs showing an audio device has noise and distortion down at -120dB or lower. Some people expect with measurements like that the SQ must be audibly perfect. Since those people expect perfection, that's what they hear, even if the sound isn't really all that perfect to an unbiased listener.
Last edited:
Do we? "Excess noise" is reserved for anything on top of the Johnson noise, in particular 1/f noise, which is not "correlated" with anything, being a fundamental property of solid state matter. It is not in particular low level (or high level, for that matter), it's just there.
So please explain your understanding of "low level signal correlated excess noise" which to me, from the physics perspective I stand on, doesn't make any sense.
So how does that make "low level signal correlated excess noise" 😀?
If you want to understand what Scott means, you may want to read this excellent paper: https://dcc.ligo.org/public/0002/T0900200/001/current_noise.pdf It shows that the noise level increases with rising voltage drop at low frequencies, following a square law, but the linearity and slope with the frequency is exactly conserved/the same.
I like this video from Amir at ASR, he explains how to do a simple ABX test at home for personnal interest and examination, he also explains how a test to "prove" anything is far more stringent and beyond pretty much everyone here. He says that results from the test he decribes can be posted on his forum as evidence that would be worthy of consideration and more investigation for underlying causes.
Audio Blind Testing - You Are Doing It Wrong! - YouTube
Audio Blind Testing - You Are Doing It Wrong! - YouTube
Ha! Explanation on ABX by a boutique audio store owner. What could possibly be the motive.
First of all, I appreciate the effort he put in it, but some important things he forgets to emphasize and sometimes he is simply wrong.
For example, the minimum number of trials needed, depends on the statistical threshold that one sets. (This threshold is in general arbitrarily choosen, for historical reasons it is quite often set at 0.05)
From a statistical point of view any combination of trials and correct results that fullfills the threshold criterion is sufficient, so it doesn't matter if someone did 5/5, 12/16 or 59/100 (first number is the correct answers, second number the trials overall), all would be sufficient, if the threshold is set at 0.05 .
Important is to train under the specific test conditions to get used to it and to use controls to check.
See for example the ABX-tests, advantage is that software tools are available , disadvantage that we already know that the protocol makes it harder for participants to reach good performance.
Further it is unlikely that any single blind test result will help you when discussing with people who demand proof. (Leaving aside that proof isn't achieveable with tests of this kind)
Even if you have done it double blind it will most likely not help, if the results are not what the are supposed to be.
Nevertheless, doing controlled listening tests (or sensory tests in general) can help to learn a lot about your perception.
But everyone should keep in mind, that fooling yourself is as easy in a "blind" listening test as it is in a "sighted" listening situation.
Yes, I agree, and is why I posted it, and I think, to a degree, what he is saying. The test he describes, which almost anyone can do, is to determine for yourself if you are really hearing a difference, not to prove anything to anyone else. I don't care if they demand proof, particularly here where we are hobbyists and enthusiasts on the whole without the resources to satisfy someone elses irrational needs.
In other words, someone says they hear such-and-such, perhaps they also supply a theory as to why, someone thinks, that's interesting or I've had an experience like that, I'll investigate if I'm really hearing that difference and perhaps find an answer. What makes me very uncomfortable is when people are shouted down when they say what they hear when someone might benefit and discover something for themselves with a little encouragement.
Last edited:
So you are making up things, as usual
. Ever heard of "voltage coefficient"? LOL, you didn't even care to open that LIGO reference.
Last edited:
If you are looking for excess noise in resistors, you may go back to carbon powder microphones used in telephones about 70yrs ago. I played with these as a child. Their noise grew with DC-current flow, this was well known and could be perceived by everyone without a double blind test. Later on carbon powder resistors showed the same effect. I remember the cheap and noisy "Vitrohm" carbon powder series. Anode resistors like these increased noise of tube amps significantly.
On the other hand voltage dependent noise of actual thin-film resistors in audio applications should be far below any golden ears resolution.
On the other hand voltage dependent noise of actual thin-film resistors in audio applications should be far below any golden ears resolution.
How would you know what the results are supposed to be?
Where can the readers look up the evidence of that?
- Status
- Not open for further replies.