I haven't always found the above analogy to apt for coupling caps. IME, sometimes a very small film bypass cap in parallel with, say, an electrolytic DC blocking cap can have some audible effect. And, no, its not necessarily imaginary. The ESR of some electrolytics is measurably different at 10kHz than it is at 100Hz. Again IME, such effects can be audible for capacitors in sensitive locations and depending on what other circuitry they are connected to.
Hi Mark,
That is due to inductance. I mentioned it in a previous post.
I'll say it again. Compare the impedance with the rest of the circuit. It does not amount to anything. You may have heard other effects, but testing with real humans over decades pretty much agrees with theory. There is nothing special or magical going on here.
If your coupling cap does develop signal voltage across it, it has become a filter element. That does matter. Otherwise this analogy does hold extremely well. That is why I used it.
That is due to inductance. I mentioned it in a previous post.
I'll say it again. Compare the impedance with the rest of the circuit. It does not amount to anything. You may have heard other effects, but testing with real humans over decades pretty much agrees with theory. There is nothing special or magical going on here.
If your coupling cap does develop signal voltage across it, it has become a filter element. That does matter. Otherwise this analogy does hold extremely well. That is why I used it.
You could always contact Chris at Parts Connexion and ask for his suggestions, they offered Signature versions of many of the Sonic Frontiers products over the years. I have always found him to be helpful.
Technical Assistance: tech@partsconneXion.com
Technical Assistance: tech@partsconneXion.com
Thanks DD, I may Just do that. I was hoping to get some insight from someone who has already preformed such action on the sfl2 and what to expect. I'm in no rush to change at all due to the pure pleasure I get from the system. As what has really been stated already it's all circuit related on what the out come will be. Expensive caps are worth every penny in the right circuit and capable system.
Not necessarily without some caveats though. Large body film caps can make good antennas. They can also sometimes be affected by squeezing them, in which case electrostriction may also affect them (shape/dimensional change due to applied potential). There are 2nd order effects that can make some expensive caps seem to sound better superficially, yet at the same time there can also be some less noticed loss of performance. For example, increased clarity with large body film caps, while at the same time some loss of soundstage depth. Depending on what is being listened for, it is possible to fool oneself into believing a large, expensive film cap is better in every way.
Last edited:
Mark you have me thinking Mundorf now. They are so well made it's even hard to tell which is the inner and outer foil. Definitely do not work as antennas.
Humble HiFi didn't test some of the caps sold by Michael Percy: https://www.percyaudio.com/Catalog.pdf
Anyway, playing with caps mostly turns out to be waste of time and money. The best possible thing is to design as many capacitors out of your system as possible. No cap sounds better than any cap, if you do it right.
Moreover some of the very best sounding, most award winning, most highly regarded audio products out there use standard industrial capacitors. Its more a question of whether the designer knows how to work with the standard parts that are commonly available.
In addition, if a cap must be used, its often best to use maybe the smallest Wima MKP2 or FKP2 that will work along with a high impedance follower, such as a well designed JFET buffer. Sure, it will have a sound, but it may be a truer sound than some giant 600v capacitor.
Anyway, playing with caps mostly turns out to be waste of time and money. The best possible thing is to design as many capacitors out of your system as possible. No cap sounds better than any cap, if you do it right.
Moreover some of the very best sounding, most award winning, most highly regarded audio products out there use standard industrial capacitors. Its more a question of whether the designer knows how to work with the standard parts that are commonly available.
In addition, if a cap must be used, its often best to use maybe the smallest Wima MKP2 or FKP2 that will work along with a high impedance follower, such as a well designed JFET buffer. Sure, it will have a sound, but it may be a truer sound than some giant 600v capacitor.
Hi bedomes,
I'm just going to say this. Chris is an honest guy, and I like him. However, never mistake that he is in the business of selling audiophile upgrade parts. The man knows how to make money and run a business.
Truth and the audiophile business is in direct conflict most of the time. I don't know if you remember Michael Elliot from Counterpoint (I was the Canadian service center until they went bust). If you look at his business model after Counterpoint went down, he sold a series of upgrades, and even gave each level a name. Posted customer orders on the web. Wrecked the units that were worked on, and it wasn't cheap! What these units actually needed was to be returned to factory stock, then the power supplies were redesigned to be quiet and reliable. That made many "bouncing" units (units that were upgraded and failed time after time) reliable enough to run without trouble - and sound better. I still fix "upgraded" units if the owners haven't given up on them.
A part is a part. It can be quantified using test equipment. Capacitors don't need break-in ... you do. We can measure a part and know how it will react in a circuit. When you measure a capacitor for example, you end up with a number of characteristics. If you apply that part to the proper circuit, it will perform well (not generate distortion or have leakage issues). Put a part that might be excellent in the wrong application and you get trouble. Distortion being the major result. But assuming they are manufactured properly, many parts are equivalent and no one can detect any sonic differences between them. We can also measure the circuit with different parts and see from the distortion spectrum (not a meter pointer!) how well that part plays with the application. For example, a polystyrene capacitor is a great part. I like them. But they don't like higher temperatures. Some don't like higher humidity. So no matter how much you like them, you avoid some applications and use a more suitable part. Very high frequency behavior may be an issue, or not. It depends.
So to mess around changing parts, you have to really know them to get where you want to go. Often that means recognizing you can't better the decision made by a team of engineers. Imagine that! So knowing the part type means reading the data sheets (not white papers), and having the means to both measure the characteristics you need, at the frequencies and circuit conditions they will face in your application. A bunch of folks agreeing with each other in ad hoc listening sessions doesn't cut it. Most of the time you don't actually hear what you think you hear. Your mind will absolutely trick you.
-Chris
I'm just going to say this. Chris is an honest guy, and I like him. However, never mistake that he is in the business of selling audiophile upgrade parts. The man knows how to make money and run a business.
Truth and the audiophile business is in direct conflict most of the time. I don't know if you remember Michael Elliot from Counterpoint (I was the Canadian service center until they went bust). If you look at his business model after Counterpoint went down, he sold a series of upgrades, and even gave each level a name. Posted customer orders on the web. Wrecked the units that were worked on, and it wasn't cheap! What these units actually needed was to be returned to factory stock, then the power supplies were redesigned to be quiet and reliable. That made many "bouncing" units (units that were upgraded and failed time after time) reliable enough to run without trouble - and sound better. I still fix "upgraded" units if the owners haven't given up on them.
A part is a part. It can be quantified using test equipment. Capacitors don't need break-in ... you do. We can measure a part and know how it will react in a circuit. When you measure a capacitor for example, you end up with a number of characteristics. If you apply that part to the proper circuit, it will perform well (not generate distortion or have leakage issues). Put a part that might be excellent in the wrong application and you get trouble. Distortion being the major result. But assuming they are manufactured properly, many parts are equivalent and no one can detect any sonic differences between them. We can also measure the circuit with different parts and see from the distortion spectrum (not a meter pointer!) how well that part plays with the application. For example, a polystyrene capacitor is a great part. I like them. But they don't like higher temperatures. Some don't like higher humidity. So no matter how much you like them, you avoid some applications and use a more suitable part. Very high frequency behavior may be an issue, or not. It depends.
So to mess around changing parts, you have to really know them to get where you want to go. Often that means recognizing you can't better the decision made by a team of engineers. Imagine that! So knowing the part type means reading the data sheets (not white papers), and having the means to both measure the characteristics you need, at the frequencies and circuit conditions they will face in your application. A bunch of folks agreeing with each other in ad hoc listening sessions doesn't cut it. Most of the time you don't actually hear what you think you hear. Your mind will absolutely trick you.
-Chris
IME there can be some of both involved. Leakage currents in new electrolytic caps may produce measurable 1/f noise. It tends to settle in to a lower leakage value over time. Probably like a lot of other things there is approximately exponential settling down into some noise floor. Most of it happens over the first hours or overnight. Some things can take a week or more. I know this to be true because I know people who have built identical products, power them up, and listened to the devices A/B with different amounts of power-on time accumulated in each one. They repeated the testing multiple times. They became convinced that some things about the sound take some time to fully settle, and that it is a property of the electronics, not of the listener.
However, in other cases I agree with @anatech that humans also tend to accommodate to the sound of an audio device over time. Absolutely. I would also agree that whatever is changing in the sound is in principle measurable, not that people are always measuring the right things. More research and knowledge has been accumulated over the years on things that are fairly easy to measure: HD/THD and noise floor. The measurements for that are almost always PSS (periodic steady state) and virtually always assume the DUT is no more that weakly non-LTI (Linear and Time-Invariant). Not everyone agrees those easy to measure and extensively studied things are all there is that is audible. Some of the people who disagree are quite accomplished in audio too. Serious disagreements among professionals. It happens sometimes.
Hi Mark,
Absolutely. Large leakage capacitors tend to form in use, but this is as you've pointed out, noise. Some folks may like a bit of tiny background noise. This only occurs when you have larger DC across the part. Even Teflon caps need to soak, but here that noise is below the rest of the audio circuits we'll see.
Your mind is your enemy in these tests. It is impossible to be completely objective no matter how hard you try. If you measure distortion and look at distortion products and noise in a spectrum, you'll find your answers. Beware environmental factors and experimental error. Your measurement needs to have an error band or accuracy envelope where you should ignore differences unless you see san actual trend. Listening depends on your mood among many other things. Even the slightest notion that something may be changing, and you will hear it. The mind.
The most sensitive electronics, like super accurate voltage references and crystals do need to settle. But again the scale is such that we poor humans could never detect the change, we simply are not that accurate and have too much "body noise" to deal with. Then there is the noise floor of the listening space. You have a very restricted dynamic range compared to instrumentation. That, and no reference at all. Doesn't sound promising - and it isn't.
Absolutely. Large leakage capacitors tend to form in use, but this is as you've pointed out, noise. Some folks may like a bit of tiny background noise. This only occurs when you have larger DC across the part. Even Teflon caps need to soak, but here that noise is below the rest of the audio circuits we'll see.
Your mind is your enemy in these tests. It is impossible to be completely objective no matter how hard you try. If you measure distortion and look at distortion products and noise in a spectrum, you'll find your answers. Beware environmental factors and experimental error. Your measurement needs to have an error band or accuracy envelope where you should ignore differences unless you see san actual trend. Listening depends on your mood among many other things. Even the slightest notion that something may be changing, and you will hear it. The mind.
The most sensitive electronics, like super accurate voltage references and crystals do need to settle. But again the scale is such that we poor humans could never detect the change, we simply are not that accurate and have too much "body noise" to deal with. Then there is the noise floor of the listening space. You have a very restricted dynamic range compared to instrumentation. That, and no reference at all. Doesn't sound promising - and it isn't.
Luckily we are not changing the design circuit here and these type's of changes most of us make is only seeing what flavours we may experience in a more expensive closer tolerance capacitor. Personally Valve amplifiers and speakers benefit the most from this, I would not bother with transistor amplifiers at all. The units In question have to be of good design to start with and then I personally will look for huge night and day improvements from the start with a pricey capacitor. If it's not there to begin with it certainly is not worth waiting around for these improvements as this would definitely cause the mind to play tricks. It really has got to be night and day positive improvements from the start on a capable system.
I honestly don't think replacing the caps with oil ones will make any audible difference. If you wish just replace with good metal film caps. They are a fraction of the price and are more reliable. Some high value ceramics can change value as function of voltage applied across them and cause distortion. So other than these and electrolytes, capacitors are usually an order of magnitude better than active components.
It is extremely difficult to get a better capacitor than a polypropylene capacitor. Teflon types are much larger, NP0 / C0G ceramic capacitors are too small. The original capacitors are about the best you can get.
High K ceramics are only good for non-critical space constrained applications.
Now, whether the circuit is solid state or tube based, it doesn't matter. Using the right capacitor or resistor is always better. I was apprenticed on tube equipment and design it - and solid state. So I see both. Close tolerance capacitors are only required as filter elements. As a pass or coupling component, it matters not within reason. Not one tiny bit. If anyone feels better about paying way too much for parts - have at it. The definition of the best coupling capacitor has nothing to do with how tight the tolerance is.
Where did price come into the quality equation anyway? All that means is the part is marked up a lot more, or is a very low volume part. What makes a good part are the characteristics, not the story a manufacturer supplies.
High K ceramics are only good for non-critical space constrained applications.
Now, whether the circuit is solid state or tube based, it doesn't matter. Using the right capacitor or resistor is always better. I was apprenticed on tube equipment and design it - and solid state. So I see both. Close tolerance capacitors are only required as filter elements. As a pass or coupling component, it matters not within reason. Not one tiny bit. If anyone feels better about paying way too much for parts - have at it. The definition of the best coupling capacitor has nothing to do with how tight the tolerance is.
Where did price come into the quality equation anyway? All that means is the part is marked up a lot more, or is a very low volume part. What makes a good part are the characteristics, not the story a manufacturer supplies.
Like always on diyA......... First post, that told you what you wanted to hear...........
To summarize @anatech ´s first post.......... You DIDN`T want to listen 🤣 🤣 🤣
Come on man I did listen honest I did. It's a lot of work to strip that pre amp down just to put four Expensive capacitors in. I just wanted to hear of other owners honest experiences . I have an anthem amp 1 that I've had since 2000 every component is factory apart from failed Bias resistors that were replaced with audio note Tantalums in 2003. I's all good. To be honest I needed to post some random crap to get started here 🙂. I'm just a spring chicken here. Anatech Makes perfect sense in the quest to educate hobbyists, That's a good thing. So calm down chaps the pre will stay in stock for some time yet😎 🤣
Tantalum Nitride? If so, we did a comparative listening session comparing that type of resistor to thin-metal-film in the output array of a DSD RTZ FIRDAC. To the listeners here where I am, the tantalum nitride resistors produced a weird sounding type of distortion (not necessarily HD/IMD) which in part sounded like 1/f noise convolved with the audio signal. To explain a bit further, convolution of the audio signal with the resistor sound (thermal noise, current noise effects which tend towards 1/f, and or possible end cap connection noise/distortion), is what happens in such a dac, as viewed in the frequency domain (it would be multiplication in the time domain view). In any case, I would not use those resistors for any purpose after that experience, except maybe for a guitar effects pedal.
Last edited: