What Moderate Priced Quality Capacitors???

[DonH50]

Depends upon where they are in the circuit, course, but probably the most-cited parameter is hysteresis. However, ESR, ESL, self-resonance, etc. all provide a measure of quality/appropriateness, as does noise, tolerance, tempco, and perhaps more important for audio, voltage coefficients. It is not too hard to measure the distortion a capacitor adds to a circuit; looking at ADC or DAC data sheets (or even regulators) will often provide a list of "good" capacitors. Note I am not primarily an audio designer, but the characteristics that matter are the same. That said, I certainly don't claim to speak for any or all audio designers; I do not have any way of knowing exactly what any designer looks at. But, good capacitors are fairly well specified, and it certainly makes sense for a good designer to take some measurements. Jung's article way back when in Audio showed that caps can matter, and measurements can help.

However, I appear to have stepped into another subjective debate so will bow out. I just don't get how these always seem to be so polarized all the way to one side or the other, when for me reality is a mix of both subjective and objective viewpoints.


Toodles - Don

 

[MylesBAstor]

Depends upon where they are in the circuit, course, but probably the most-cited parameter is hysteresis. However, ESR, ESL, self-resonance, etc. all provide a measure of quality/appropriateness, as does noise, tolerance, tempco, and perhaps more important for audio, voltage coefficients. It is not too hard to measure the distortion a capacitor adds to a circuit; looking at ADC or DAC data sheets (or even regulators) will often provide a list of "good" capacitors. Note I am not primarily an audio designer, but the characteristics that matter are the same. That said, I certainly don't claim to speak for any or all audio designers; I do not have any way of knowing exactly what any designer looks at. But, good capacitors are fairly well specified, and it certainly makes sense for a good designer to take some measurements. Jung's article way back when in Audio showed that caps can matter, and measurements can help.

However, I appear to have stepped into another subjective debate so will bow out. I just don't get how these always seem to be so polarized all the way to one side or the other, when for me reality is a mix of both subjective and objective viewpoints.


Toodles - Don

You know Tom it's that nonsense that continues to be promulgated that high-end designers are idiots, know nothing about circuit design and have never done a measurement in their life! It's never ceases to amaze me how there are people here who know more than the equipment's actual designer and will redesign it for them. Browse the LAMM website for one designer who's both proud of his measurements and of his sound.

As far as the nonsense about voicing goes, riddle me this: Which part do you select if two, three or more components measure the same? And I'll bet you the top resistors, capacitors, etc. measure very closely (Hint: take a look at the drug industry ).

 

[DonH50]

Was that directed at me or Tom? Either way seems a bit over the top... I certainly never said high-end designers are idiots; known/met far too many sharp designers. I am not sure what "nonsense about voicing" I spoke but I am clearly out of my league in this discussion.

 

[microstrip]

Depends upon where they are in the circuit, course, but probably the most-cited parameter is hysteresis. However, ESR, ESL, self-resonance, etc. all provide a measure of quality/appropriateness, as does noise, tolerance, tempco, and perhaps more important for audio, voltage coefficients. It is not too hard to measure the distortion a capacitor adds to a circuit; looking at ADC or DAC data sheets (or even regulators) will often provide a list of "good" capacitors. Note I am not primarily an audio designer, but the characteristics that matter are the same. That said, I certainly don't claim to speak for any or all audio designers; I do not have any way of knowing exactly what any designer looks at. But, good capacitors are fairly well specified, and it certainly makes sense for a good designer to take some measurements. Jung's article way back when in Audio showed that caps can matter, and measurements can help.

However, I appear to have stepped into another subjective debate so will bow out. I just don't get how these always seem to be so polarized all the way to one side or the other, when for me reality is a mix of both subjective and objective viewpoints.


Toodles - Don

Don,

Thanks for an excellent post. I see you refer that voltage coefficients could be an important issue in capacitor influence in sound quality. However I only know about work carried about them in ceramic capacitors, that no audio designer would use in the signal path. Do you have any reference about a good work about this subject in film capacitors?

IMHSO (in my humble subjective opinion), the reason why capacitor sound always becomes a polarized subject, even between people who accept this effect, is mainly because we are not able to establish a correlation between electrical properties and sound quality, except when there is a clear degradation. We are still trying to extrapolate from excellent, but now old work, and forget that the claims made about new capacitors exceed the scope of our references. No one at that time addressed the mechanical stability, mechanical self resonances, the effect of foil thickness or/and the metal foils composition and thickness. Until there are clear references about these subjects the debate will always become polarized, as no bridge is possible between the subjective and the objective.

 

[KBK]

Hello to you microstrip,

I am familiar with the famous "Mundorf" capacitors. I think they are in the very elite class of caps with few others being able to come up to their level. I will be looking to see if they can be had for reasonable prices. Thanks much sir.

jazzbo bob

The lower priced mundorf caps are an excellent series of compromises and design choices, as all capacitors actually are. That mundorf has turned out an excellent product, for the funds involved.

 

[KBK]

Don,
My main question is not philosophy, but a practical question - what measurement do designers use to select their signal capacitors for sound quality? Except for those capacitors that have very poor characteristics and no one uses in the signal path, is there any measurement that correlates with sound quality?

I don't look at just specs, I look at the mechanical build vs the materials in the molecular sense. For example, in a given film cap, what is the metallic foil deposition method and what is the density of the deposition layer, how was it deposited. What is the qualitative origin point of the film in use, and so on. Then, the build proper, and thus the complex field considerations in the context of propagation, and self noise/motor issues. All of these things need to be viewed as a set, and some of it will show up in both measurements and listening.

The most interesting thing is to always remember that at the peak of any transient, motion is zero, thus a full collapse of the field and then return of the field in the opposite polarity is assured.

That skin effect and complex body (of the capacitor) originated full field distortions occur in the peak moment of any micro or macro transient. Not only once, but twice. The collapse and then the re-establishment of complex field. Twice.

When you add in the point that the ear works only with the micro and macro-transients in order to decode the signal..then the major distortion parameters of the given capacitor are the singular doorway through which the ear/brain works, then you've got a recipe for the so called 'low level micro-distortions' of capacitors (the distortions are called minor in engineering terms and weighting of measurements)..being very 'hearable' by audiophiles.

That the ear/brain, is ONLY working in the area where so called micro distortions occur. that 100% of what we hear, is the area where engineering weighed and originated measurements define as 'minor' or miniscule, as compared to the whole signal. Nice engineering terms.

But..the ear does not care..... nor does it hear/decode that way.

In the most literal sense possible, the ear is only listening in the tiny area in which distortions occur in the propagation of AC complex electrical fields as they 'work' (do their thing) or as they are realized in all materials used in a piece of audio gear.

That in a wire, or a capacitor, or transformer, or transistor or a valve and so on..that the operational area in which the signal distortions originate...that the ear is ONLY listening to those areas of distortion origins, due to how the ear functions.

 

[ack]

Was that directed at me or Tom? Either way seems a bit over the top... I certainly never said high-end designers are idiots; known/met far too many sharp designers. I am not sure what "nonsense about voicing" I spoke but I am clearly out of my league in this discussion.

I think the response was directed at Tom and incorrectly quoted you. I agree with your assessment. Finally, I think the John Curl document Tomelex is referring to may be http://www.q-audio.com/johncurl.pdf - search for capacitor and you will find a lot of relevant information, like:

it is important that I have done research on capacitor distortion and have published it, especially with regards to tantalum coupling caps.

There are many imperfections in capacitors. DA is related to the material in the dielectric. Aluminum electrolytic caps tend to have the worst DA. Dissipation factor is only slightly related to DA, but is most concerned with heating of the cap, due to its losses. This 72 includes DA, but also connection resistance and lead resistance. DF is important in switching supplies so that the filter cap does not overheat and explode. ESR is again related, but is mostly concerned with the 'short-circuit' aspects of the cap, and how well it will pass signals without any drop across it. These are approximations, and you all can quibble with me about them, but it is important to understand that they are not all the same thing.
DA was found to be important, first, in analog computers, popular in the 1950's. It was known to effect calculations, and was difficult to compensate for. It is also very important in sample and hold circuit operations. Tantalum and ceramic caps can have lots of nonlinear distortion as well. Aluminum caps are better, but not perfect in this respect... 'Smear' is a pretty good description of DA. I once used the term 'echo' for DA effects... I recommend looking at Photo #6 of a 10uf aluminum cap compared to an 8 uf polypropylene cap differentially subtracted with a (nominal) 50k load on both caps in the article: 'A Real-Time Signal Test for Capacitor Quality'.

we (Walt Jung, Scott Wurcer, and I) found that even a single capacitor can have a measurable residual when directly compared to another capacitor of similar value, but slightly different material composition

Section 5.4 is dedicated to capacitors...

 

[MylesBAstor]

Was that directed at me or Tom? Either way seems a bit over the top... I certainly never said high-end designers are idiots; known/met far too many sharp designers. I am not sure what "nonsense about voicing" I spoke but I am clearly out of my league in this discussion.

Don this wasn't directed at you.

 

[MylesBAstor]

I don't look at just specs, I look at the mechanical build vs the materials in the molecular sense. For example, in a given film cap, what is the metallic foil deposition method and what is the density of the deposition layer, how was it deposited. What is the qualitative origin point of the film in use, and so on. Then, the build proper, and thus the complex field considerations in the context of propagation, and self noise/motor issues. All of these things need to be viewed as a set, and some of it will show up in both measurements and listening.

The most interesting thing is to always remember that at the peak of any transient, motion is zero, thus a full collapse of the field and then return of the field in the opposite polarity is assured.

That skin effect and complex body (of the capacitor) originated full field distortions occur in the peak moment of any micro or macro transient. Not only once, but twice. The collapse and then the re-establishment of complex field. Twice.

When you add in the point that the ear works only with the micro and macro-transients in order to decode the signal..then the major distortion parameters of the given capacitor are the singular doorway through which the ear/brain works, then you've got a recipe for the so called 'low level micro-distortions' of capacitors (the distortions are called minor in engineering terms and weighting of measurements)..being very 'hearable' by audiophiles.

That the ear/brain, is ONLY working in the area where so called micro distortions occur. that 100% of what we hear, is the area where engineering weighed and originated measurements define as 'minor' or miniscule, as compared to the whole signal. Nice engineering terms.

But..the ear does not care..... nor does it hear/decode that way.

In the most literal sense possible, the ear is only listening in the tiny area in which distortions occur in the propagation of AC complex electrical fields as they 'work' (do their thing) or as they are realized in all materials used in a piece of audio gear.

That in a wire, or a capacitor, or transformer, or transistor or a valve and so on..that the operational area in which the signal distortions originate...that the ear is ONLY listening to those areas of distortion origins, due to how the ear functions.

How about polymer orientation? Plastics such as polystyrene for instance come in many "grades" and are used for many purposes. What might be optimal for optical clarity in say a petri dish (polymer chains all oriented in the same direction) may not be the right choice say for a cap where you don't want the cap to hold onto the signal and disorganized polymeric chains may be a better option.

 

[KBK]

How about polymer orientation? Plastics such as polystyrene for instance come in many "grades" and are used for many purposes. What might be optimal for optical clarity in say a petri dish (polymer chains all oriented in the same direction) may not be the right choice say for a cap where you don't want the cap to hold onto the signal and disorganized polymeric chains may be a better option.

Yes, and that the draw and healing of the plastics over the given wire can account for some of it's perceived directionality. To test the directionality of the given wire, it has to be stripped of it's skin, then checked..then checked as a finished item, with skin, and then the two states compared.

it can't be the same piece of wire on both tests, to some degree, as stripping the wire of it's skin requires a play time to occur, to some degree, due to nano and micro breakage in the copper or element/alloy in use -from the handling required to strip the skin.

 

[DonH50]

Most of the capacitor data I have gleaned through the years has been from manufacturer data (working directly with them to get more than in the usual datasheets) and measurements. Some of the measurements, especially at RF, require a significant investment in resources (equipment and people, like $1M+ in equipment plus fancy test jigs and many manweeks of development and test time). Things like hysteresis and mechanical stability are a challenge to measure, especially when you need a range of test cases over numerous corners (process, termperature, large and small signal voltage, etc.) Simple pulse testing can be revealing but requires a 'scope or test system capable of extremely fine detail on a large step. I used a special test rig for most of those measurements, fast (ns - ps) diodes to clamp and precision supplies to prevent overdriving the 'scope (otherwise it will tend to dominate the step response). Mechanical testing I was only peripherally involved with and it was a nightmare as one might imagine.

IMO the measurements can be taken to readily show differences among capacitors, but having the resources and knowledge to do them is not at all trivial. Most do a small subset of measurements and then (especially in the audio world,at least IME) just try them out. I strongly suspect prototype design and build variation tends to mask many effects. And we are stil left with the issue of correlating measured differences to what we hear, again IME often nontrivial. Like many things audio, I think it is less "can we measure" than "is it practical" and would such data be made public...

All IMO - Don

p.s. DA is responsible for a number of ills, including (mainly) hysteresis and noise.

 

[opus111]

And we are stil left with the issue of correlating measured differences to what we hear, again IME often nontrivial. Like many things audio, I think it is less "can we measure" than "is it practical" and would such data be made public...

I have a hypothesis that a large chunk of the SQ differences due to caps might well be down to 'incidental' factors like how effective the cap is as an antenna (most audiophile boutique caps are huge, in electronics terms) and also the resonances that they produce at RF frequencies. Large physical size means large parasitic inductance. In my designs, smaller physical size equates to better hence I'll use NP0/C0G and I don't hear any colouration from these at all. I'll even use the very lowly regarded X5Rs for coupling, again I haven't noticed any detrimental effects so far but I do keep an open mind that they might be degrading the sound in ways I have yet to notice.

 

[stellavox]

I have a hypothesis that a large chunk of the SQ differences due to caps might well be down to 'incidental' factors like how effective the cap is as an antenna (most audiophile boutique caps are huge, in electronics terms) and also the resonances that they produce at RF frequencies. Large physical size means large parasitic inductance. In my designs, smaller physical size equates to better hence I'll use NP0/C0G and I don't hear any colouration from these at all. I'll even use the very lowly regarded X5Rs for coupling, again I haven't noticed any detrimental effects so far but I do keep an open mind that they might be degrading the sound in ways I have yet to notice.

Interesting point Opus 111,

However one would hope that the designer has taken into account what I will refer to more generally as "outside grunge" when "actualizing" the enclosure to house the design. Your hi fi equipment is subjected to and CAN GENERATE electric, magnetic and electro-magnetic fields. I've previously opined that some "tweaks" involving metal objects placed around the room may actually act as a shunt or short circuit for some of this grunge, resulting in better sound because they reduce the effect of grunge on some sensitive part of your system.

Some manufacturers actually send their equipment to an independent facility where it can be put in a cell and subjected to "electric/magnetic hell", and see how it responds (I came across a test report on a piece of Cello equipment). As you can imagine, this costs a LOT of money. Then there's mechanical vibration - think "my product is housed in a 25lb billet of machined unobtanium...??" Oops, forgot cryogenically treated! The result is that it probably DOES "sound better"!

Charles

 

[DonH50]

I have a hypothesis that a large chunk of the SQ differences due to caps might well be down to 'incidental' factors like how effective the cap is as an antenna (most audiophile boutique caps are huge, in electronics terms) and also the resonances that they produce at RF frequencies. Large physical size means large parasitic inductance. In my designs, smaller physical size equates to better hence I'll use NP0/C0G and I don't hear any colouration from these at all. I'll even use the very lowly regarded X5Rs for coupling, again I haven't noticed any detrimental effects so far but I do keep an open mind that they might be degrading the sound in ways I have yet to notice.

Hmmm... I have a hard time seeing how RF resonances impact the sound unless they are large and somehow aliased or rectified back into the audio band, or so bad they impact the performance of the buffer amps. Large physical size does not have to mean large inductance, and may mean less since leads and conductive films are larger -- look at RF transmit caps. That said, there has been so much focus on making small, high-quality capacitors (and everything else) that I would not argue your general experience. In fact it matches mine; physically smaller caps do in general have wider frequency response, albeit also lower values. Note in virtually all cases the self-resonant frequency is well above the audio band; I agree at the output of a DAC you want that self-resonant peak to be outside the glitch frequency so 10's MHz or more is not at all out of line.

X5R is notorious for having high voltage nonlinearity and value changes with bias, as well as being somewhat microphonic; your experience might argue against capacitors having a significant impact on sound.

Very interested in following your work, opus111!

 

[opus111]

I have the slight advantage here in that I have done the job of electronics design engineer for an audio company, I did it for around 8 years. So I know firsthand its true that audio designers are idiots, in the sense that I've learned a huge amount since I quit the industry I have also taken the equipment I've designed for EMC testing (its required for CE marking normally) and eventually it has passed so have a reasonable handle on what's involved. Yes it is expensive, particularly if you don't pass first time.

I do though doubt that 'shorting out' radio waves outside the enclosures can have much effect

 

[opus111]

Hmmm... I have a hard time seeing how RF resonances impact the sound unless they are large and somehow aliased or rectified back into the audio band, or so bad they impact the performance of the buffer amps.

I tend to think that the former effect is happening - RF intermodulating with the audio. I put up a paper as an attachment in my 'Digital sounds like analog' thread just today which deals with this. How I came across it started with the fact that I found opamps in my active speakers sensitive to seemingly minor layout differences and figured this must be due to RF effects. Adding inductors before opamp inputs helped improve the sound a lot as did adopting rigorous star earthing.

X5R is notorious for having high voltage nonlinearity and value changes with bias, as well as being somewhat microphonic; your experience might argue against capacitors having a significant impact on sound.

Yes I am most certainly aware of the hideous voltage coeff of these capacitors, so much so that I deliberately oversize (in terms of value, not physical size) them by a factor of 10 or more. But when 4.7uF is available in 0603 size this isn't too much of a problem - 6 in parallel still takes up almost no space at all, this is the beauty of working in SMT

Very interested in following your work, opus111!

Thanks for the encouragement Don!

 

[DonH50]

I realized that in your case you have a lot of unshielded boards (not in a case) so RF interference/EMI could indeed be an issue, especially using wideband opamps. The fact that ferrites (beads, inductors) helped is a clue.

 

[stellavox]

Full metal shielding will work, but it has to be very extensive, like think of it as essentially being "waterproof" and it gets the job done. Like is done in real world cryptographic (secret) installations of which I had some little bit of experience. However, stopping it in circuit is the most logical route with audio, as you are keenly aware.

Right Tomelex, BUT then you cut holes in the box and attach "antennas" like speaker cables - or an interconnect say to a phono cartridge that eventually may leave a "shielded" environment.

Charles

 

[MylesBAstor]

Right Tomelex, BUT then you cut holes in the box and attach "antennas" like speaker cables - or an interconnect say to a phono cartridge that eventually may leave a "shielded" environment.

Charles

Geez Charlie you're such a party pooper!

 

[Jazzbo]

Thanks everyone for this great feedback. This is all welcomed information.

jazzbo