 
|
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 |
68.13.78.187
In Reply to: This could only work for low frequencies, posted by Al Sekela on October 29, 2006 at 12:12:21:
Audio frequencies are pretty low. I think the cable can be seen as a lumped parameter for those frequencies. For instance, if a signal travels at 1/2 c, and the highest frequency of interest is 100 kHz, a 10 degree phase difference is .93 mi * 5280 ft is about 5000 ft, en degrees of which is about 160 ft. I think 160 ft for one degree at a frequency 5 * higher than we can hear is pretty ignorable.Another concern is the group delay for the follower or the op amp. That could be managed by engineering in similar group delays.
Follow Ups:
Audio wavelengths are indeed so long that all practical audio cables can be treated as lumped elements. Unfortunately, RF noise of certain frequencies is amplified by low-loss cables that act as resonators (for which the cable length is an integer multiple of the half-wave), and most audio equipment is unable to avoid intermodulation of RF noise with the audio signal. This makes the RF properties of audio cables important contributors to their sonic performance, and, I believe, helps explain the mysterious "system synergy" we keep reading about. It makes the driven guard scheme impractical for high quality audio applications.
I follow everything you say except your conclusion. To accept the conclusion seems to require an additional tenet: That system synergy, or voicing, or euphonic distortion is desireable. If you accept that, which I think can be alternately stated as, "We can't get rid of distortions so let's balance them," then your position makes sense. But that doesn't get us closer to science replacing art.I prefer the position of eliminating distortions, which points out the remaining problems that can then be addressed. I think active cables will get us a little closer to that goal. It also makes long cable lengths easy.
Audio cables have RF properties that can be controlled to reduce their influence on the sound. IME, the sound is better if the RF issues are reduced without introducing other problems. However, your proposal will result in uncontrolled RF problems, for basic reasons already explained.You seem to have misunderstood John Curl's reply. He is talking about a DC bias on the cable's dielectric, not a driven guard scheme.
If there is a problem with "switching" polarities on dielectrics, then anything that uses a coupling capacitor with a non-biased source and feeds a grid sitting at zero bias (like a cathode biased stage) will generate that kind of distortion. That's a typical topology for phono stages, which get amplified unmercifully. So one would presume that any problem with that topology would be readily apparent in such a stage. Has anyone heard that kind of problem? I would need to see a physics-based explanation on why switching polarities might pose a problem.On the other hand, switching polarities across dissimilar metals I can see as a problem, and that takes place all the time. Every dissimilar junction acts like a tiny bi-directional diode, in that the voltage has to exceed a small threshold before significant conduction occurs. That should be a problem, and one that can be overcome with dc bias, but I've never seen it addressed.
at the interface with the insulating dielectric to avoid metal-semiconductor junction voltage offsets at the signal zero-crossings (and consequent zero-crossing offsets in capacitance). I believe you've answered your own question as to why a DC bias on an audio cable's insulation can make a difference in performance. There is no way to keep the copper from corroding from atmospheric sulfates, as there is no practical dielectric insulation that is impervious to gas diffusion. The diode effect at the interface will cause a capacitance offset at the signal zero crossing. A DC bias is a simple way around this issue.
I think that what Al and others are saying is, that we don't have a perfect pre-amp, how can you postulate a perfect driver amp?If the driver amp ever goes 180 degrees with the signal amp OR breaks into oscillation or other mis-behavior, this will make things worse, not better.
Your response seems to be that if it were only designed properly, it wouldn't do this. But by the same token, if the original preamp circuit "were designed properly", it wouldn't need the help of the driver amp to neutralize the cable capacitance.
Now, in theory, it should be possible to create a driver amp and signal amp pair with deliberately controlled bandwidth and softly limited drive capability, so that the driver amp would roll-off before it got into trouble, and would not be able to oscillate, etc. Then, even if it committed errors in terms of driving the inner 'shield', they would still be way down compared to the signal amp driving the raw cable with it's full capacitance. One could liken it to to a 1st order neutralization of the cable capacitance, with both the signal and driver amps tapering off gently in the higher frequencies, and well before either can get into RF problems. Of course, the bandwidth has now been limited, and the cable itself could still pick up RF energy, which could still negatively impact the signal amp OR the driver amp, etc.
Thus, even a well thought out scenario could still, and probably WOULD be prone to problems and errors in the cable capacitance neutralization.
I think that this is what other folks have been trying to point out, but I could be way off base.
From another angle, I think that trying to neutralize the cable capacitance is going down the wrong road. WHY are we trying to neutralize the cable capacitance in the first place? Typically because of the load it places on the signal amp, and the subsequent distortions due to the current demand on transients, etc. In other words, we want to use the driver amp because the signal amp is less than perfect. The driver amp will be less than perfect too, and so, it will not perfectly cancel the cable capacitance.
However, resistive isolation of the signal amp from the cable capacitance can be quite effective in greatly reducing the need for a high level of current drive, and while this does reduce bandwidth, we were going to have to do that for the driven scenario anyway in order to make a practical real-world cirucit work at all.
The thing is, the cable dielectric will be precariously 'balanced' near zero volts, with the error residue being the driving force for causing it to flip back and forth from one polarity to the other. This may in fact cause MORE overall total signal distortion than letting the signal amp drive the cable and deal with the (now reduced with that last resistively isolated scenario) current demands.
If we want to try and reduce or eliminate the effects of the cable dielectric from impinging on the signal, it would seem that a better method would be to use some form of dielectric bias, ala Audioquest DBS and others
For some DIY methods, see: http://www.audioasylum.com/audio/cables/messages/95641.html
and: http://www.audioasylum.com/audio/cables/messages/124517.htmlThis way, you keep the dielectric from ever flipping it's overall polarity, it is operated 'single-ended'.
Jon Risch
Good point, Jon, about biasing the dielectric. I haven't tried it myself, but if it could reduce dielectric absorption, or change it is some good way, then I could go for DC polarization. Actually, as you know, there are practical examples of driven shields and DC polarization of cables out there, and both have had some success.
It seemed like a good idea, and if you say some folks have had some success, I think I'll get around to trying it.
Why would you want that? In polymers commonly used for dielectrics, and at temperatures below the glass transition, the major mechanism of dielectric polarization is the polarizability of the chains, i.e., the slight shift of electron density toward the positive potential.
Which has been hypothesized as the mechanism for DA effects.Let's look at it another way.
With a zero bias dielectric, the signal is having to charge the dielectric from say +2 V peak to -2 V peak. The current demands and the DA time constants are kicking in big time every time the signal swaps polarity.
Now, place a 72 volt bias on the dielectric, now the dielectric is being asked to go from 70 volts to 74 volts. We go from a 200% change in dielectric potential through zero, to a 5.5% change and staying at the same polarity.
Jon Risch
Which has been hypothesized as the mechanism for DA effects.
If I'm not mistaken, I believe it's space charge that's responsible for DA, which I don't think is what MKJ is talking about here.
With a zero bias dielectric, the signal is having to charge the dielectric from say +2 V peak to -2 V peak. The current demands and the DA time constants are kicking in big time every time the signal swaps polarity.
Current demands? What are you saying, that biasing the cable reduces its capacitance?
Now, place a 72 volt bias on the dielectric, now the dielectric is being asked to go from 70 volts to 74 volts. We go from a 200% change in dielectric potential through zero, to a 5.5% change and staying at the same polarity.
Seems to me that either way, the dielectric's still being polarized by +/- 2 volts.
se
From another angle, I think that trying to neutralize the cable capacitance is going down the wrong road. WHY are we trying to neutralize the cable capacitance in the first place? Typically because of the load it places on the signal amp, and the subsequent distortions due to the current demand on transients, etc.
More typically it's done for stability reasons.
However, resistive isolation of the signal amp from the cable capacitance can be quite effective in greatly reducing the need for a high level of current drive, and while this does reduce bandwidth, we were going to have to do that for the driven scenario anyway in order to make a practical real-world cirucit work at all.
RL isolation is effective at isolating the signal amp from cable capacitance at RF while maintaining sub-ohm impedance at audio frequencies, negating the reduction of bandwidth from resistive isolation.
The thing is, the cable dielectric will be precariously 'balanced' near zero volts, with the error residue being the driving force for causing it to flip back and forth from one polarity to the other. This may in fact cause MORE overall total signal distortion than letting the signal amp drive the cable and deal with the (now reduced with that last resistively isolated scenario) current demands.If we want to try and reduce or eliminate the effects of the cable dielectric from impinging on the signal, it would seem that a better method would be to use some form of dielectric bias, ala Audioquest DBS and others...
This way, you keep the dielectric from ever flipping it's overall polarity, it is operated 'single-ended'.
So far I haven't seen so much as a shred of evidence, or even a halfway plausible theory as to why typical dielectrics would behave their worst about the zero point. That's rather like saying a transformer would perform its best with a DC "biasing" current flowing through it that's significantly greater than the signal current.
se
With that kind of 'pre-load', I see little point in even attempting to discuss this with you. I mean, RL on a PREAMP?????? Yeeuch!
Jon Risch
With that kind of 'pre-load', I see little point in even attempting to discuss this with you. I mean, RL on a PREAMP?????? Yeeuch!
What exactly is the problem, Jon? As I said, in the audio band its impedance would be less than an ohm. But would be substantially more at RF, isolating the preamp from the cable capacitance and negating the use of a higher value single resistor.
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 ]
