I think listening and measuring form two poles of an iterative process where the two inform and feed into each other, but at the end of the day, the determining question is "how does it sound?"
A good scientist takes seriously the approximate nature of any current scientific theory (the uncertainty of objectivity). This attitude implies an openness to data that doesn't fit the prevailing regime. In audio, this data perforce includes listening.
If the shoe doesnt fit... well, that's good for you. My comments were not directed to anyone in particular. I worked all my adult life in high-level R&D and coordinated many large scale projects, high visibility projects for DOD and DOE with some of the brightest minds in existance and I Know for sure they F**k up real bad at times. Hundred-million dollar programs have been halted. Lots of horror stories. So, my advise for audio engineers is based on real practical experience. I was told, LLNL did not hire theororist... they would go to a university if they needed more theory. It was an applied science lab. It was often bad theory or assumptions from design engineers and physicist that got things blown up. I mean blown up - in some cases, literally. And, this is from the Best minds. The rest of us mortals (me too) in face of this become humble and skepticle but at the same time open for new ideas to explore that can explain something. Your advise has been noted and considered. -RNM
Last edited:
Bad use of "Distorts" - this is still not a non-linear distortion mechanism in a otherwise linear circuit app, only an unfamiliar type of frequency dependence that requires a continuous model - like a lossy transmission line - no proposed part of any shown models are nonlinear in the sense of generating new frequency components from a excitation signal
there is dielectric hysterisis, you can find papers on it in "ferroelectric" caps, - I've never seen it quantitifed in "good" film dielectrics
Last edited:
His name is Alex Zaslavsky (alexanderzas in Skype). Say Hello from me.
No distortion here just fractional calculus to make a better continuous model. I've known Ken since he was a grad student and my wife worked for him as an intern, as usual good work. Since you persist in being wedded to these mis-conceptions I am wasteing my time here, talk about closed minded. Giving up on them is too threatening, where have I heard that before? This is getting as bad as the recurring Matti Otala saga.
Last edited:
Signal-level distortion itself depends on the capacitor's distortion characteristic. Two ceramic caps of the relative same dielectric constant can have markedly different distortion characteristics. Compare the following two measurements from Bateman of two capacitors having the same or similar size, ESL/ESR, but a different dielectric material. The first is of an X7R ceramic, the second of a C0G ceramic:
Attachments
Well that's just silly, Scott. The model attempts to mathematize how capacitor functioning deviates from that of an ideal capacitor. You can call that deviation a fractional calculus. Regardless what you call it, it distorts AC signal transmission through the capacitor. And as we've agreed (?), capacitors "function" only on AC.
I understand S.W.'s comments on many things which are being described as distortion which, technically, it isnt from the model(s). many people describe what they hear as distortion when something has got worse in some way. yes, it isnt always "real" distortion. Although in some case it is. So, I try to use the term distortion in a broad way... not only by math and engineering terms. Sometimes, the affect of something is AS IF or EQUIVELANT to the affect of harmonics, resonances or nonlinearities. So, thats why I call these things How Linear Distortion can Sound Non- linear and tests to bridge the gap. -RNM
Last edited:
Of course, that is obvious. You saidserengetiplains said:
which is not true. The level of distortion depends critically on the application.
To clarify still further: a particular capacitor may have a non-linear dielectric with a second-order (parabolic) response to applied voltage. Such a capacitor will generate 3rd-order distortion if run with an AC signal and no DC bias - the distortion will vary like the cube of the signal voltage. If it has a DC bias it will in addition generate 2nd-order distortion, proportional to the DC bias and the square of the signal voltage. So the basic dielectric response is a property of the capacitor, but the type and level of distortion will depend on the application.
There are minor frequency response deviations in this model as jcx says no new frequency components. If you chose to call (for instance) an ideal 2Hz high pass a distortion of the input signal the word loses meaning and clear technical discussion is hampered. Common use is for components that do not have a linear relationship between the state variables associated with it.
Humor me take some data sheets and compute the magnitude of these issues in real applications, Bateman's data is certainly not directly applicable to power supply bypass or input coupling at frequencies far above the pole. He takes most of his data with large AC voltage on the capacitor in order to more easily see anything.
Last I remember storing charge is a useful "function" too.
Last edited:
I agree with the first part of your sentence and disagree with the second, if I'm reading it right. Yes, application influences type and level of distortion, but the type and level of distortion a given application produces will itself be influenced by the distortion properties of the dielectric.
In other words, dielectric distortion exists no matter the application. Thus for the same application, two capacitors otherwise identical or sufficiently similar in all other respects but their dielectric will produce different levels of distortion referable, therefore, only to the dielectric. Dielectric-induced distortion is an irreducible, non-dependent quality of the chemical makeup of the dielectric. In a given application, that dielectric will, yes, vary in its functioning or operation dependent on the application, but that functioning will carry and reproduce the distortion characteristic of the dielectric concerned.
Yes it is, and it is that very function---charge storage---that DA attempts to measure. Capacitors operate because they store and release charge in response to varying voltage on their terminals. But they don't store and release cleanly. The "residual" release can be viewed as a spurious voltage injected into the circuit in question. This release is semi-active, in my view, and seems to me somewhat unique in electronic components.
Thank you. At last you have admitted what you were previously denying. I don't know whether you have changed your mind, or are now expressing yourself more clearly.serengetiplains said:
You still appear to be confused though:
No. Dielectric non-linearity exists no matter the application. The distortion produced by that non-linearity depends critically on the application, and the effect of that distortion also depend on the application. In the same way, amplifier non-linearity exists whatever the signal but distortion only occurs if the signal is big enough to invoke the non-linearity. I think you are confusing non-linearity with distortion. The former is the cause of the latter when combined with signal level.
Btw, for those who haven't read or want to again read Bateman's articles on capacitors, here's a link where you can find them:
Pro Audio Design Forum • View topic - Cyril Bateman Capacitor Sound Archive
Pro Audio Design Forum • View topic - Cyril Bateman Capacitor Sound Archive
Fine, but take that list of dielectrics, Teflon, pp, .... and make a 2Hz input DC blocking network then take some measurements in a typical set up at normal listening levels or even line level if you want (20Hz and up). We are not talking dB's of errors, and actually quantifying things often weakens the arguments. Ed's resistor measurements are all well and good but I can't get excited about -110db vs -130dB.
Maybe it's time to revive the in-amp cap null fixture since we now have high resolution measurement capabilities that we did not have then.
I haven't changed my mind, and I can see how some of my wording can be taken differently than I intended.
And I can accept your distinction between distortion and dielectric non-linearity. I've only wanted in my posts to point to that non-linearity as a function of the dielectric itself. I called it distortion, much like Richard understood above, in reference to a capacitor's real v. ideal functioning, but referencing dielectric effects only (DA).
And one further point for clarity, yes, a capacitor's entire functioning, including its ESL and ESR, must be fitted to the application in question. That a capacitor's DA generally correlates with its dielectric constant represents a problem for circuit fidelity, if I can use that phrase, because a better performing capacitor may be inappropriate or ineffectual given its size. What to do in that case?
I can only take the "wording" you use, not what you intended to say. I did try to clarify your meaning, but you persisted in saying something which you now admit you didn't intend to say.
DA and dielectric non-linearity are two completely different and unrelated issues. Please don't confuse them. There is enough confusion in this thread already! It is very difficult having a technical discussion when people say A causes X, when they may actually mean to say that B causes Y, and the actual situation in reality may be that C causes Z!
It is well past midnight in the UK so I am off to bed!
DA and dielectric non-linearity are two completely different and unrelated issues. Please don't confuse them. There is enough confusion in this thread already! It is very difficult having a technical discussion when people say A causes X, when they may actually mean to say that B causes Y, and the actual situation in reality may be that C causes Z!
It is well past midnight in the UK so I am off to bed!
I told in Linear Audio what my threshold was back when I had my greatest hearing capability. -comparing -110 to -130 with sine waves probably cant be heard. BUT, is there a test that could indicate audible levels? i have shown one possibility.
I never did a listening test at such low distortion levels. I am not the source of saying such a test IS audible. i have just put out the info and others then looked for the 'best' quality cap to use and then marketing took over to 'educate' the public. At least so it seems. Thus, all cap types became bad except the 'best' or most ideal. That's my view from the bridge of my nose. -RNM
I never did a listening test at such low distortion levels. I am not the source of saying such a test IS audible. i have just put out the info and others then looked for the 'best' quality cap to use and then marketing took over to 'educate' the public. At least so it seems. Thus, all cap types became bad except the 'best' or most ideal. That's my view from the bridge of my nose. -RNM
Last edited:
- Status
- Not open for further replies.