|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
128.125.179.204
In Reply to: RE: Motor run capacitors revisited posted by Tre' on May 07, 2017 at 12:48:44
It is similar to transistor distortion. A typical Best Buy transistor amplifier has distortion of less than 0.1%, compared to 5% of a 2A3 amplifier. But which one sounds better?
For motor run capacitors, tgδ of 0.005 at 1,000 Hz means that signal losses at this frequency are 0.5%. This is so little that it can be ignored for practical purposes, as long as insertion loss of a capacitor is concerned. But insertion loss figure tells nothing about distortion.
Looking closer at the phenomenon of dielectric absorption (aka dielectric memory), which is largely responsible for capacitor losses, one could see that influence on the signal may be not so benign. To demonstrate dielectric absorption, a capacitor is charged to, say 100 V, then shorted for 10 sec. 10 sec after the short is removed, residual voltage on the capacitor is measured. In most organic dielectric capacitors, this residual voltage is 2-5% of initial value, and in electrolytes it could be as high as 15%. Not small.
At the molecular level, polar dielectrics (e.g. vegetable oil in motor run capacitors) consist of molecules with positive charge on one end and negative on the other, or dipoles. When voltage is applied to capacitor, dipoles orient themselves in electric field. When voltage polarity changes, dipoles re-orient. Re-orientation takes time, causing dielectric hysteresis (similar to magnetic hysteresis). Effects of hysteresis increase with frequency. How much distortion is caused by dielectric hysteresis, I don't know, but one sure thing is that this type of distortion increases with frequency.
Examples of polar dielectrics: paper, mylar, PCBs, mica, some ceramics, vegetable oil.
Non-polar dielectrics: polypropylene, polystyrene, teflon, mineral oil.
The difference between polar and non-polar dielectrics is quantitative rather than qualitative. All substances have positive and negative parts in their structure. Only vacuum is true non-polar dielectric.
Follow Ups:
...You mention DA as a possible source of "capacitor distortion". I've wondered about this because the theoretical model of DA is relatively simple and should be very easy to simulate in spice or maybe even wire up and measure...at least at a high level. And yet, at this point I've never seen this done WRT audio apps. DA is a known issue with certain analog circuits like S&H and integrators but I'm not sure what's known transfers well to audio.
FWIW, the non-electrolytic caps that exhibit the greatest DA using my somewhat crude technique are the Russian K40s, esp after exposure to temperatures of say 150F. For this reason I've been reluctant to use them.
I think the simple part is dielectric constant ε, which is somewhat related to dielectric absorption. ε is a coefficient indicating how many times capacitance is increased by a given material compared to vacuum. Materials with high dielectric absorption usually have high ε.
Dielectric absorption and permitivity may not be easily susceptible to modeling. In the idealized situation, the relaxation time parameter is used, which is time needed for elemental dipoles of dielectric to re-orient upon instantaneous change of electric field. The assumptions are linear, homogeneous, isotropic materials with uniform response to changes in electric field. Real situation is far from these assumptions. Dielectric memory, as capacitor shorting experiment shows, persists for a very long time (seconds), indicating that relaxation times in real materials may have wide spread. Plastic films are highly anisotropic. How linear and uniform dipole response to applied electric field is, only God knows. Modeling has been successfully used to predict such gross phenomena as losses or capacitance changes with frequency (BTW, another source of capacitor non-linearity), but I doubt its usefulness in calculating distortion.
...I guess that's my point; lots of doubt and speculation regarding DA effects in audio but no actual testing or even modeling. The fact that high DA electrolytics have been (relatively) successfully employed in audio causes one to doubt its importance but doubt there is.
The high level model of DA I've seen is a network of series RC networks paralleled across the ideal C. The time constants of the networks are a range of many seconds to many tens of seconds. Even if not an entirely accurate model, it ought to useful in determining whether or not DA is indeed something of audible concern. If nothing else, it should be easy to model and examine for gross high level effects.
Perhaps the difference in capacitors is in fact measurable, but not represented as frequency response, THD or RMS intermodulation. Transient distortion in amplifiers is a neglected subject, and I am highly suspicious that it accounts for much of what we hear that is otherwise impossible to explain.
--------------------------
Buy Chinese. Bury freedom.
Alexander Pushkin, "Mozart and Salieri". One of my favorite audiophile citations.
Dielectric absorption is only one source of capacitor distortion. There are several others. One is capacitance change with frequency, and one more is capacitance change with voltage. Distortion due to capacitance change may be quite serious - any change in capacitance causes changes of voltage across the capacitor. Yet one more is capacitors acting as mechanotransducers, which I posted on a while ago. This was about capacitors acting as low quality electrostatic speakers and low quality condenser microphones at the same time.
Maybe capacitor distortion has been studied, and we just don't know? However, it seems more like there is no interest to this subject in the field of electronic engineering. We'll have to do it ourselves if we want to know. But even if we find that capacitor distortion is low, who is going to be convinced? I know from experience that some capacitors sound better than other, and avoiding capacitors in the signal path is a good practice. People whose judgment I trust preferred the sound of transformer- or direct-coupled amplifiers to that of capacitor-coupled amplifiers.
"it seems more like there is no interest to this subject in the field of electronic engineering."
I posted here on this exact subject earlier this year. We (at my day job) had just discovered that the capacitance of modern multi-layer capacitors changes markedly with voltage. So much so that the capacitors - for bypassing purposes - virtually disappear. Only a few manufacturers of these components are publishing the relevant data.
--------------------------
Buy Chinese. Bury freedom.
.
Have Fun and Enjoy the Music
"Still Working the Problem"
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: