 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
Back in the Jitter fiasco thread, you made a comment regarding Pease's DA model that's been bouncing around in the back of my mind since then. You said:I also personally believe that Bob's DA model is incomplete for audio purposes, and does not include a set of RC networks with diodes in them, to account for polar dielectric behavior.
Pease's model is fundamentally no different than the model used by Jung & Marsh. I'm just wondering exactly what it is you're trying to model that you would model with a diode. You say "polar dielectric behavior." Can you be more specific? It seems like you might be referring to an electret, but I don't see that being modeled with diodes. So what is it you are in fact referring to?
se
Follow Ups:
As John alluded to in his post, I am talking about a 1st order approximation of the non-linear portion of a dielectric that is NOT as linear as most film caps are. Really poor mylars, electrolytics, etc. might need this additional circuit element to more completely model the behavior of these caps to a higher level of resolution.Yes, I am aware that this is as far as the current model goes, and that Jung and Marsh also used it to model DA effects.
I am proposing an additional RC circuit similar to those already there, but one with a diode in series with the RC. In order to more completely model some real world capacitors, there might need to be several of these, with different time constants, and different diode polarities.
Pease's simple two section DA circuit compensation only reduced the DA effects by a factor of ten (and you can see where a portion of the correction was overdone initially, due to only two sections and trying to adjust for a centerline cancellation), wherin he compared this reduction of a polypropylene cap to the DA quality level of a teflon cap. If we think in terms of a teflon capacitor may still be considered to be audibile compared to no capacitor, or at the least, that a middle of the road polypropylene is audible, then modeling down to levels well below Pease's percentages would be necessary.
Jon Risch
As John alluded to in his post, I am talking about a 1st order approximation of the non-linear portion of a dielectric that is NOT as linear as most film caps are. Really poor mylars, electrolytics, etc. might need this additional circuit element to more completely model the behavior of these caps to a higher level of resolution.Yes. But exactly what non-linear portion is that? You said originally that the diodes were "to account for polar dielectric behavior." And that's what I was looking for something more specific on. Again, it seems as if you might be referring to electret properties but it's not entirely clear.
I am proposing an additional RC circuit similar to those already there, but one with a diode in series with the RC. In order to more completely model some real world capacitors, there might need to be several of these, with different time constants, and different diode polarities.
Well that's the thing. You need to know just which property it is that you're trying to model so you can produce a model that will reasonably behave like that property. And what I'm unclear on is just which property it is that you're wanting to model. I mean, just because you're aware of a general non-linearity, doesn't mean that throwing some diodes into the mix will necessarily model that particular non-linearity with any degree of accuracy. See what I'm getting at?
Pease's simple two section DA circuit compensation only reduced the DA effects by a factor of ten (and you can see where a portion of the correction was overdone initially, due to only two sections and trying to adjust for a centerline cancellation), wherin he compared this reduction of a polypropylene cap to the DA quality level of a teflon cap. If we think in terms of a teflon capacitor may still be considered to be audibile compared to no capacitor, or at the least, that a middle of the road polypropylene is audible, then modeling down to levels well below Pease's percentages would be necessary.
Yes. But I think there's the assumption here that DA is what's responsible for the non-linearities you're wanting to model and I'm not sure that's the case. DA seems to be pretty linear, both measured and modeled. Which is why I was wondering why you seemed to be wanting to introduce non-linear diodes into a property that by all accounts seems to be quite linear.
That's why I asked you if you might be referring to electret properties when you mentioned "polar dielectric behavior" as any electret properties would introduce some degree of non-linearity. Dipole hysteresis would be another which comes to mind. But I don't see either of these being effectively modeled with diodes.
se
se
Jon Risch, I also have been looking at this problem for a long time. I would think that the model normally used for dielectric absorption was made for polystyrene caps back in the 1950's. Walt Jung and I have papers going back that far. It does NOT describe any non-linear distortion, and polystyrene usually doesn't have much non-linear distortion in any case. Of course, when you change over to mylar (polar?) or electrolytic caps, the model still works for the linear distortion, but I would agree that you would have to add a non-linear part to the model to get the non-linear part in the model. I have used diodes in series with resistors in the past, to model non-linear pots, etc. I can see that it could work for caps, to a first order at least.
Of course, when you change over to mylar (polar?) or electrolytic caps, the model still works for the linear distortion, but I would agree that you would have to add a non-linear part to the model to get the non-linear part in the model.Why must there necessarily be a non-linear part to the DA model? The non-linearities may well be due to properties other than DA. And I suspect that might well be the case as the linear DA model gives a very good fit to empirical data.
There may be other things going on beside DA which are accounting for capacitor non-linearities. Electret properties and dipole hysteresis are two obvious non-linarities that come to mind. Until factors such as these are also considered, you might end up running down a blind DA alley as far as looking for the source(s) of non-linearity in capacitors is concerned. And until you identify the source(s), you're not going to be able to model them adequately.
se
You may not know this Steve, but Walt Jung and I spent a year on this problem, and published an article on DA in caps. Later, 'High Fi News' published our article and Martin Collums, Ben Duncan and Dr. Hawksford followed up on our work.
24 years ago, I gave a paper to the IEEE 'Audio and ... ' section, including cap nonlinear distortion of both tantalum and ceramic caps. I've done the measurements and the cap modeling, and we even footnoted Bob P's paper in our later publication. We all know who Bob Pease and Doug Self are, and how they feel about audio quality. I also know that they tend to ignore the effects of DA on audio quality.
I have also seen measurements of aluminum caps that showed non-linear distortion, measured by Black Gate more than 15 years ago.
If we only have passive devices to work with in modeling a cap, diodes might be the only effective way to construct a non-linear distortion model. If Jon Risch has any ideas on this subject, I would like to see his input. Of course, the non-linear distortion will most probably fall below the linear distortion in magnitude, so for DA itself, the linear model is probably OK, but it is NOT the complete model of a capacitor. I suspect that your concern is the fact that Jon mentioned a 'polar' dielectric. What this means is that the plastic polarizes with electric signal and both stores and dissipates electrical energy. I have also seen a 'non-return to zero' distortion in monolythic ceramics that probably needs a battery to emulate, but these caps have so much non-linear distortion, a very accurate model will not be necessary. This is why I did not show this distortion in my early paper for the IEEE in 1978.
You may not know this Steve, but Walt Jung and I spent a year on this problem, and published an article on DA in caps. Later, 'High Fi News' published our article and Martin Collums, Ben Duncan and Dr. Hawksford followed up on our work.
24 years ago, I gave a paper to the IEEE 'Audio and ... ' section, including cap nonlinear distortion of both tantalum and ceramic caps. I've done the measurements and the cap modeling, and we even footnoted Bob P's paper in our later publication. We all know who Bob Pease and Doug Self are, and how they feel about audio quality. I also know that they tend to ignore the effects of DA on audio quality.
I have also seen measurements of aluminum caps that showed non-linear distortion, measured by Black Gate more than 15 years ago.Uh, so? What's that to do with anything I said and how does it preclude other more obvious sources of non-linearities in capacitors?
If we only have passive devices to work with in modeling a cap, diodes might be the only effective way to construct a non-linear distortion model.
Why would you only have passive devices to work with? For that matter, why would you necessarily need any devices at all except perhaps as an aid in visualization? A model is simply a mathematical construct. It doesn't require any particular kind of devices.
If Jon Risch has any ideas on this subject, I would like to see his input.
As would I. Which is why I asked him to expand on his original comment.
Of course, the non-linear distortion will most probably fall below the linear distortion in magnitude, so for DA itself, the linear model is probably OK, but it is NOT the complete model of a capacitor.
Of course it's not the complete model of a capacitor. That was my point.
I suspect that your concern is the fact that Jon mentioned a 'polar' dielectric.
I haven't any particular "concern." I just wanted to know exactly what Jon meant when he said "polar dielectric behavior" so I wouldn't have to try and second guess him.
What this means is that the plastic polarizes with electric signal and both stores and dissipates electrical energy.
Or doesn't polarize with the electric signal as the case may be. The phrase Jon used is a bit ambiguous as several inferences could be drawn from it. That's why I asked Jon specifically what he was referring to.
I have also seen a 'non-return to zero' distortion in monolythic ceramics that probably needs a battery to emulate, but these caps have so much non-linear distortion, a very accurate model will not be necessary.
Great. That means more play time for us. :)
se
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2013, Audio Asylum, all rights reserved ]
[ Take5 ]
