Impersonating a Mod is an offence. I suggest you apologise to them and plead ignorance.
OK, now I understand why you seemingly plucked this poor design from the air and proclaimed it a good example to follow, so supporting the idea that special caps are needed for audio: you already built a version of it.
Whether omitting the output coupling caps is a good idea would depend on the likely output offset voltage. Or perhaps by "suppress" you did not mean 'omit'?
Anyway, this is getting away from the topic of this thread. No need to continue discussing a poor TI design.
OK, now I understand why you seemingly plucked this poor design from the air and proclaimed it a good example to follow, so supporting the idea that special caps are needed for audio: you already built a version of it.
Whether omitting the output coupling caps is a good idea would depend on the likely output offset voltage. Or perhaps by "suppress" you did not mean 'omit'?
Anyway, this is getting away from the topic of this thread. No need to continue discussing a poor TI design.
silverprout - A reminder to read the Forum Rules.An ESR measurement is a measurement of the real component of impedance in a device (as opposed to the imaginary component, which is a reactance). This real component might be due to the component having a purely linear series resistance, or it might be due to some sort of phaseless loss mechanism that looks just like resistance because it has no phase associated with it to make it appear reactive, but is in fact a result of nonlinearities.
So, while high ESR does not guarantee a nonlinear dielectric, low ESR does preclude all sorts of real losses.
As an example, I measured a number of inductors for their distortion. I found that in general, the inductors that exhibited more distortion had a greater real component of impedance than those that had less distortion. The interesting part with inductors is that generally, the ESR can be specifically known, separately from the real component of impedance, since they are wound with specific sized conductors, whose actual resistance can be known.
So, yes, this doesn't directly have anything to do with capacitors, but it shows that 'ESR' can be a garbage bin for the real component of any and all sorts of losses, which may be nonlinear or plain old resistance.
Therefore, choosing 'low ESR' is not always just to minimize resistance, it's to minimize other loss mechanisms that manifest as phaseless losses, and are therefore hard to distinguish from resistive losses.
The ESR (or DF) is often quoted in manufacturer's datasheets, and this gives at least a little handle on comparing quality. According to xtronics (https://xtronics.com/wiki/Capacitors_and_ESR.html), "ESR rating of a capacitor is inversely related to its quality."
I have never seen a datasheet that states distortion specs for capacitors. There doesn't seem to be anything else, other than ESR (or DF) for comparison. For instance, a very good 2.2uF electrolytic capacitor will have an ESR component of ~2.5% its impedance at 1 KHz; whereas a polypropylene cap will have over 20 times less. That's enough of a difference to be significant.
I have never seen a datasheet that states distortion specs for capacitors. There doesn't seem to be anything else, other than ESR (or DF) for comparison. For instance, a very good 2.2uF electrolytic capacitor will have an ESR component of ~2.5% its impedance at 1 KHz; whereas a polypropylene cap will have over 20 times less. That's enough of a difference to be significant.
Last edited:
The series of articles from Cyril Bateman ( https://linearaudio.nl/cyril-batemans-capacitor-sound-articles and https://linearaudio.net/volumes/2221 ) give a pretty good idea of how much what type of capacitor distorts, but indeed, datasheet figures usually don't tell you anything about distortion. The only exceptions I know are some ceramic class-2 capacitor datasheets that have a capacitance versus voltage graph, which shows you quite clearly that these should not be used as audio DC blocking capacitors.
The claim that ESR is inversely related to quality is true by definition when by quality you mean the quality factor or Q factor, Q = abs(Xc)/ESR. Whether that has anything to do with audio quality is a different matter.
In any case, I think that Monte McGuire and you are correct about the correlation between ESR and distortion. Typically the capacitor types that have the least ESR also have the least distortion, the least dielectric absorption and the most acceptance from audiophiles.
Thanks for the link....interesting reading, but you would think a large corporation like WIMA could at least afford to hire someone whose FIRST language is English to edit their papers.
Yes, but I guess the point is, at what point does the distortion become inaudible? My current dilemma is choosing between a 3.3 uF polypropylene cap and a 3.3 uF polyester cap. The polypropylene cap, of course, has 10 times LESS ESR than the polyester, which has 10 times less ESR than an electrolytic. But the polypropylene is also 6 times bigger than either the polyester or the electrolytic.
The document should be diffused in a "publishing" format... as PDF file for exemple.
The word format is only for editing purposes, we don't see what occur in companies, but i stay dubitaive when intelligent people are doing s.......
uch things all together.So, my english is not perfect and the document writing seems good to me, the layout of this "High End" paper is definitely on the "Low End" side IMHO.
I've found that document by searching on google a meta-information that i've found in the original PDF file from the Wima site.
By just making a right click on the properties/PDF information/title
You will see this :
"Test Audio-ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo"
Yes, my error. And to be totally fair, I have found electrolytics that are only 8 times more ESR (DF=0.08). So, electrolytics 8 times more ESR than a polyester, which is ten times more than a polypropylene. So an electrolytic is 80 times the ESR of a polypro.
Thanks, appreciate your thoughts.
Are there any particular components that a larger film cap would be likely to interact with in terms of stray capacitance (to try to avoid proximity to), or is it more circuit than component dependent?
And are the HF losses (and damping) directly related to lower ESR on a polyester - thus the same case applies for a polypropylene?
I thought it might be better posting the schematic marking the capacitors in question along with a photo showing the orientation, this is probably an extreme case since the replacements were polypropylene (Panasonic ECWF) and notably larger.
Schematic shows (4) parts for one channel, the 2.2uF 25 x1 and 50V x1 (both BP) are for the DC servo, and 2.2uF 100V x2 (polar) are general filtering from my understanding.
I can't notice any difference subjectively (when those parts alone were changed) but am curious if you see any obvious theoretical flaws.
Attachments
Two main things to look out for when thinking about stray capacitance are:
1. AC mains (or heater supply) on nearby conductors - mostly a problem at high impedance points in the circuit - the effect is increased hum
2. another circuit point with significant size (e.g. another cap?) which carries the same signal with the same phase but very different amplitude - scope for positive feedback leading to instability or at least an HF peak
Polypropylene or polyester or any other film cap is likely to have low ESR so the same issue arises if the original circuit designer was relying on electrolytic ESR (either deliberately or accidentally) to dampen a resonance.
In many cases the only way to find out (for both issues mentioned in this post) is to try it. Sometimes it works OK, sometimes it does not. The main point is that naive 'upgrades' can sometimes do harm. Be aware that a small amount of signal degradation (e.g. HF peaking falling short of oscillation) may be heard but could be misperceived as an improvement - especially if lots of money has been spent.
1. AC mains (or heater supply) on nearby conductors - mostly a problem at high impedance points in the circuit - the effect is increased hum
2. another circuit point with significant size (e.g. another cap?) which carries the same signal with the same phase but very different amplitude - scope for positive feedback leading to instability or at least an HF peak
Polypropylene or polyester or any other film cap is likely to have low ESR so the same issue arises if the original circuit designer was relying on electrolytic ESR (either deliberately or accidentally) to dampen a resonance.
In many cases the only way to find out (for both issues mentioned in this post) is to try it. Sometimes it works OK, sometimes it does not. The main point is that naive 'upgrades' can sometimes do harm. Be aware that a small amount of signal degradation (e.g. HF peaking falling short of oscillation) may be heard but could be misperceived as an improvement - especially if lots of money has been spent.
Old types of voltage regulators sometimes rely on electrolytic capacitor ESR; I know from experience that an LM317 with only large ceramic output decoupling capacitors can become a triangle oscillator. Connect a run-of-the-mill electrolytic capacitor with higher value in parallel with the ceramic or film caps and the LM317 is stable again.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.