Which Capacitors for Audio?

[JohnRoberts]

I did a fair amount of Philosophy of Science back when I was at university. It's the subject that intellectually justifies the claims that science can advance knowledge, while also pointing out that it can never make a claim that something is true. In the 40 years since, I have never seen ANY scientific endeavour that looks anything like this intellectual framework. Actual science, or practical science, doesn't remotely adhere to what it claims is its intellectual justification. As an example, I will be impressed if you can find me a single audio company that regularly uses double blind listening tests to assess its products. There are plenty of companies that have tried, but admitted they either uncovered them or gave up on them because is was unfruitful. You need people who know what they are listening for and who have good ears. That comes by exposure and through experience and they are few and far between.

Incidentally, science does not require belief. Religions do, and in this context that's called scientism, which is a disease of science.

I don't know of any major companies who design by ear... Human hearing is too unreliable for important work.

Audio phoolery is closer to religion than hard science .

JR

 

[abbey road d enfer]

Double blind tests are somewhat similar to substractive tests. They allow putting in evidence differences but not what is "best".

 

[Newmarket]

IIRC, they were CDC documents. I can't remember if I saved them but they're probably not that hard to find on the net. In this country (UK) we know officially that each doctor was paid an additional £12.50 for each patient vaccinated and that at least £50 or £60 billion was added to the NHS budget. The NAO has probably published the breakdown of where that money was spent by now, and if so it will be easily found on its website. But you prefer to call me a liar. Thanks for that.

Much more of this and I won't be on this website.

I'm not particularly interested in the USA figures on this. Probably says more about the structure of healthcare / pharma in the USA rather than anything specifically related to test methodology. And I didn't say you actually lied but you said yourself that it was 'likely' ie you don't know it to be true (wrt £250 000 per doctor in UK).
Your figures just don't work out. At £12.50 each a doctor would need to vaccinate 20 000 times to see £250 000. Seems a tad unlikely when the average number of patients per GP is significantly less than 2000. Even if you allowed for people getting multiple vaccinations over a couple of years. And overlook the arguable impracticality of time availability. And of course that money would be subject to taxation at the highest rate (ignoring tax advantaged investments / savings etc) so they wouldn't see all of that income with a big chunk of it going back to HMRC.
Oh, and of course more vaccinations were given by non doctor staff / volunteers at vaccination centres and pharmacies etc.

 

[abbey road d enfer]

Remember that before this the usual speaker cable (considered good enough because people thought it made no difference) was either mains cable (and probably not even 1.5mm2) or bell wire.

I started my professional life in audio in 1973, worked for a HiFi manufacturer, when they still existed in my country. I've never seen one of our amps or speakers connected with less than 1.5mm². On the contrary, I've seen often 2.5mm² and sometimes bigger, which necessitated ferrules or spades. Our brochures were very clear about the necessity of good speaker cables.

 

[abbey road d enfer]

Cable may be 0.1 ohm nowadays but it didn't used to be until people started taking cable seriously! Quite a lot of things to do with the Q have been ignored in the past and several still are today.

The main justification for promoting high damping factor is loudspeaker control, so it's been taken seriously since the early 50's.

I've been looking at this quite recently and my estimate is that quite a large number of speakers that were sized for, say, a Q of 0.707 actually have a Q closer to 1.

For a time, the magic number for loudspeaker Q was 0.383, because it's the one that gives a true B'worth response. The quest for smaller speakers lead to decreasing Q, so many speakers now have a Q of less than 0.3. A speaker with a Q of 1 is a lemon. Even 0.707 is not desirable.

I have tried to find these rises in temperature and I haven't been able to. I seem to remember Keith Howard also tried at one time too, after me, and couldn't either. Your estimate, if it comes from what you've read, may be based on the work KEF did on the T27 and B110 (maybe in an LS3/5a,

It's not an "estimate", it's the result of many measurements I did. Voice-coil temp measured with a DC CCS (Constant Current Source), magnet temp, cabinet temp and ambient measured by thermocouples. Signal was a sum of 1/3 octave spaced tones, cause pink noise made DC readings too difficult, even with strong filtering.
The resonant frequency and Q vary by as much as 10% in an overnight test at 50% of rated power.
Goal was assessing power handling for concert sound reinforcement.
Of course domestic speakers are used differently, but even the difference between a winter day and a summer day changes the Q and resonant frequency much more than any variation of decent cable.

 

[molke]

Double blind tests are somewhat similar to substractive tests. They allow putting in evidence differences but not what is "best".

Doesn’t that depend on the actual test? Difference is just one goal of listening tests. I could imagine tests (e.g. using a semantic differential) where the goal is to find out what sounds best. For statements that can be generalized to some degree (in contrast to subjective listening), the effort is quite high (as already mentioned).

 

[ct.waveform]

I don't know of any major companies who design by ear... Human hearing is too unreliable for important work.

Audio phoolery is closer to religion than hard science .

JR

Where did I say, or even suggest, the products are designed by ear? But one would hope that every company listens to its products before releasing them onto the market, so as to assess whether the design process has met its hopes and expectations. The minimum for this would be that it is at least better than the product it's replacing and hopefully better than its competitors. Do you know of ANY company that uses double blind testing to make this assessment and relies solely on its results? I don't even know one that would use it at all.

You are a Moderator and representative of this website. Correct? I have seen some of your moderating and the less said about that, the better. Are veiled insults part of the required moderating skills here? So far, I'm in need of a dislike button, am a liar, and am now an audiophool. This is not looking promising. I'm not fussed by the audiophool accusation and am usually happy to defend it, but coming from a moderator is different.

While I remember, scientific "truth" does not have a statistical component. Something does not become more likely to be true the more tests it passes. Karl Popper covers this at some considerable length in Conjectures and Refutations. Nor is science a consensus endeavour.

 

[abbey road d enfer]

Blind testing with audio often doesn't yield the best results either, because differences can be very small, specific and your ability to judge can change drastically over - say - the course of a day. You may easily prefer the - ultimately - worse option at a given moment.

Double-blind studies were originally devised to assess the validity of certain choices when there is an objective measure of the results. Primarily medical treatments.
Extending them to subjects where no objective measurement exists limits the validity to assessing if there is a perceptible difference or not.
Determination of "best" is left to the old method of polling.
That's what Apple Harman have done with their earbuds and headphones surveys. The conclusion is that the poll results influence the design of headphones and earbuds, which in turn influences the polls and that feeds a vicious circle. One of the results is that a notable proportion (40% IIRC) of the youth are going to be hearing impaired.

 

[living sounds]

I do imagine that ears played a big role in developing many of the most beloved and idiosyncratic pieces of gear, be it preamps, synthesizers or reverb units. Lucky accidents maybe, but kept because subjectively the result was preferable.

 

[living sounds]

Double-blind studies were originally devised to assess the validity of certain choices when there is an objective measure of the results. Primarily medical treatments.
Extending them to subjects where no objective measurement exists limits the validity to assessing if there is a perceptible difference or not.
Determination of "best" is left to the old method of polling.
That's what Apple have done with their earbuds and headphones surveys. The conclusion is that the poll results influence the design of headphones and earbuds, which in turn influences the polls and that feeds a vicious circle. One of teh results is that a notable proportion (40% IIRC) of the youth are going to be hearing impaired.

Indeed.

 

[abbey road d enfer]

I do imagine that ears played a big role in developing many of the most beloved and idiosyncratic pieces of gear, be it preamps, synthesizers or reverb units. Lucky accidents maybe, but kept because subjectively the result was preferable.

No doubt about it, but the development always start with a set of objective requirements.
"Sounding good" is implicit from the start, but there's nothing in the arsenal of a designer that translates that requirement in topology, component values, bias point, whatnot.
Subjective evaluation always come after many initial choices have been settled; very seldom a product is completely redesigned from scratch after subjective evaluation.
Some may call it designer's instinct, but it's more a case of using experience and applying proven methods.
The first Fender amps came from a valve manufacturer's cookbook.

 

[abbey road d enfer]

Doesn’t that depend on the actual test? Difference is just one goal of listening tests. I could imagine tests (e.g. using a semantic differential) where the goal is to find out what sounds best.

Then who/what is going to decide which one sounds best? Of course there are cases where the dfferences are so gross anyone can tell, but we are talking of subtleties like the diference between a 1.5 and a 2.5mm² cable. I know that the results in a double-blind test are gonna be 50/50. So yes there may be a difference, but no winner.
I've witnessed cases where the winner in such a competition had objective flaws.

 

[Matador]

Then who/what is going to decide which one sounds best? Of course there are cases where the dfferences are so gross anyone can tell, but we are talking of subtleties like the diference between a 1.5 and a 2.5mm² cable. I know that the results in a double-blind test are gonna be 50/50. So yes there may be a difference, but no winner.
I've witnessed cases where the winner in such a competition had objective flaws.

In one of my guitar amp prototypes I had mistakenly used 0.1u coupling caps as I had mistakenly placed them in a pile of visually-identical 0.047uF bin of other caps. I had delivered it to a tester who wanted to use it at a practice session, and I had replaced those caps based on my own testing (originals were 0.022uF). I was in a rush, and didn't play-test it after the cap switch.

When I got it back, he told me he really loved the tone and to change nothing, but when I plugged it in, I hated the tone. WAY to much bass response, way to thick and muddy in the high-bass / low mids, which was opposite of the original problem I had with it was that it was too thin so I thought to increase the coupling caps but just a bit. But I obviously over-shot it by nearly two times.

So who's to say who's right?

 

[JohnRoberts]

Where did I say, or even suggest, the products are designed by ear?

As an example, I will be impressed if you can find me a single audio company that regularly uses double blind listening tests to assess its products. There are plenty of companies that have tried, but admitted they either uncovered them or gave up on them because is was unfruitful. You need people who know what they are listening for and who have good ears. That comes by exposure and through experience and they are few and far between.

Expert ears?

Back in the 70s/80s designing studio effects (Loft delay/flanger) it was pretty much a wholly design by ear exercise. Since then my focus is on clean linear audio paths. I let the bench test numbers tell the story.

But one would hope that every company listens to its products before releasing them onto the market, so as to assess whether the design process has met its hopes and expectations. The minimum for this would be that it is at least better than the product it's replacing and hopefully better than its competitors.

I have written about this right here for years (decades). My last day job at major manufacturer we had QA people in the factory performing crude listening tests right on the production lines. Humans can be pretty good at finding deviation from normal in that environment. QA people discovered a noise problem from a newly approved capacitor vendor. Engineering had already approved the new capacitor but production QA discovers a noise issue. I never got to the bottom of the "problem", I just black-balled the new vendor and switched back to quiet capacitors. ( I suspect the new caps were not properly formed in, but didn't waste any more of my time on them).

I have long opined that listing tests serve a function to confirm that we have actually measured the correct factors. I determined decades ago that I can measure stuff that I cannot hear, but so far I have always been able to measure things that I can hear. So listening tests play a role backing up bench testing.

Do you know of ANY company that uses double blind testing to make this assessment and relies solely on its results? I don't even know one that would use it at all.

No... I have used plenty of single blind tests, not perfectly controlled, almost as much to capture a group sentiment for subjective sounding circuits like clip limiters (DDT) and the like. Results were not always the expected or desired. For example the unsophisticated listeners often preferred the sound of amplifiers allowed to clip (most likely they preferred the somewhat louder, higher power output).

You are a Moderator and representative of this website. Correct?

I am an unpaid volunteer moderator. My personal opinions are mine alone.

I have seen some of your moderating and the less said about that, the better. Are veiled insults part of the required moderating skills here? So far, I'm in need of a dislike button, am a liar, and am now an audiophool. This is not looking promising. I'm not fussed by the audiophool accusation and am usually happy to defend it, but coming from a moderator is different.

I have a very long history dealing with the esoteric audio community, try not to take it personally I am not trying to insult you, but making perhaps a knee-jerk response. FWIW "jerk" is part of knee-jerk, mea culpa.

While I remember, scientific "truth" does not have a statistical component.

that actually sounds true.

Truth is more of a philosophical construct. But I am not a philosopher so will stop now.

Something does not become more likely to be true the more tests it passes. Karl Popper covers this at some considerable length in Conjectures and Refutations. Nor is science a consensus endeavour.

Not to change the subject but modern climate science appears to be very much a "consensus endeavor".

Sorry about the veer... my personal opinion again.

JR

 

[ct.waveform]

The main justification for promoting high damping factor is loudspeaker control, so it's been taken seriously since the early 50's.

Damping factor is a near meaningless figure. As Doug Self says in one of his early books, the amplifier does not have an iron grip on the loudspeaker. Not anything like. Halving or doubling the damping factor will usually make virtually no difference to the response.

For a time, the magic number for loudspeaker Q was 0.383, because it's the one that gives a true B'worth response. The quest for smaller speakers lead to decreasing Q, so many speakers now have a Q of less than 0.3. A speaker with a Q of 1 is a lemon. Even 0.707 is not desirable.

This is completely wrong. A Q of 0.707 is the maximally flat Butterworth response. Get any table of filter shapes - there's one in Walt Jung's Op Amp Handbook which is online - and you'll see the Q of a 2nd Order Butterworth is 0.707. There are almost no speakers in the world with a Q of 0.3. Even in free air, precious few speaker driver units will have a Q as low as 0.3. As you reduce the cabinet size for a given driver the Q goes up, not down. I suggest you look at some data sheets from any reputable manufacturer and learn how to calculate the Q and resonant frequency for a given cabinet size from the VAS, which is the equivalent volume of air for that driver (or a driver with the same cone area). A Q of 1 means it has an output of 1 at the resonant frequency. There is not much wrong with that, though it does have a slight rise inside the pass band (I can't remember exactly what, but it's somewhere around 1.5dB). A Q of 0.707 is probably historically the most desirable target response of all time because it has the flattest amplitude response. There are others that people have aimed at for different reasons, such as a Q of 1/2 because it has no overshoot or a Q of 0.56 because it has a Bessel characteristic.

It's not an "estimate", it's the result of many measurements I did. Voice-coil temp measured with a DC CCS (Constant Current Source), magnet temp, cabinet temp and ambient measured by thermocouples. Signal was a sum of 1/3 octave spaced tones, cause pink noise made DC readings too difficult, even with strong filtering.

Testing to near destruction with a non-music signal which doesn't have the spectral content of music can hardly be said to be representative of normal domestic use.

The resonant frequency and Q vary by as much as 10% in an overnight test at 50% of rated power.

Did you discover that the resonant frequency and Q vary with amplitude, too? Both getting lower and so compensating somewhat for the rise in resistance. This is one of the things that sets that Goldilocks level for listening. Which reminds me that you could go to Klippel.de for a primer in how loudspeakers behave at low frequencies. It's a truly excellent site.

Goal was assessing power handling for concert sound reinforcement.

Which is miles away from my goal of making sure the transfer functions stay close to the knee of the target slopes in normal use.

Of course domestic speakers are used differently, but even the difference between a winter day and a summer day changes the Q and resonant frequency much more than any variation of decent cable.

Noticing first that your idea of a decent cable is what has only become commonplace since cables became a hi fi product in their own right, I think your own figures might dispute that. You got a change of compliance in the spider and the less dense air of only 20% with the voice coil and magnet probably running at a steady temperature of over 100 degC. Unless your speakers are outdoors in Scandanavia, I doubt the ambient temperature variation would be very much more than 10C, so the change in Fs and Q would be a touch over 1%. Moreover, they would likely reach pretty much the same steady state temperatures inside the cabinet in either season.

Still, this is all getting rather cantankerous so I shall have a quick look at the notification I have and this will be my last post on this site.

 

[JohnRoberts]

Damping factor is a near meaningless figure. As Doug Self says in one of his early books, the amplifier does not have an iron grip on the loudspeaker. Not anything like. Halving or doubling the damping factor will usually make virtually no difference to the response.

Yes... damping factor was meaningful for old school tube amplifiers because of their relatively high output impedance. It was pretty much meaningless for modern solid state amps until a few decades ago when some marketing puke at Crown started advertising their amplifier's high numerical DF to create a merchantable difference between otherwise similar amps. Other amp makers including my employer responded in kind for marketing purposes. (Peavey patented an amp circuit capable of negative source impedance.)

This is completely wrong. A Q of 0.707 is the maximally flat Butterworth response. Get any table of filter shapes - there's one in Walt Jung's Op Amp Handbook which is online - and you'll see the Q of a 2nd Order Butterworth is 0.707. There are almost no speakers in the world with a Q of 0.3. Even in free air, precious few speaker driver units will have a Q as low as 0.3. As you reduce the cabinet size for a given driver the Q goes up, not down. I suggest you look at some data sheets from any reputable manufacturer and learn how to calculate the Q and resonant frequency for a given cabinet size from the VAS, which is the equivalent volume of air for that driver (or a driver with the same cone area). A Q of 1 means it has an output of 1 at the resonant frequency. There is not much wrong with that, though it does have a slight rise inside the pass band (I can't remember exactly what, but it's somewhere around 1.5dB). A Q of 0.707 is probably historically the most desirable target response of all time because it has the flattest amplitude response. There are others that people have aimed at for different reasons, such as a Q of 1/2 because it has no overshoot or a Q of 0.56 because it has a Bessel characteristic.

Testing to near destruction with a non-music signal which doesn't have the spectral content of music can hardly be said to be representative of normal domestic use.

reliability and ruggedness to survive whatever customers throw at gear is a useful feature.

Did you discover that the resonant frequency and Q vary with amplitude, too? Both getting lower and so compensating somewhat for the rise in resistance. This is one of the things that sets that Goldilocks level for listening. Which reminds me that you could go to Klippel.de for a primer in how loudspeakers behave at low frequencies. It's a truly excellent site.

Which is miles away from my goal of making sure the transfer functions stay close to the knee of the target slopes in normal use.

Noticing first that your idea of a decent cable is what has only become commonplace since cables became a hi fi product in their own right, I think your own figures might dispute that. You got a change of compliance in the spider and the less dense air of only 20% with the voice coil and magnet probably running at a steady temperature of over 100 degC. Unless your speakers are outdoors in Scandanavia, I doubt the ambient temperature variation would be very much more than 10C, so the change in Fs and Q would be a touch over 1%. Moreover, they would likely reach pretty much the same steady state temperatures inside the cabinet in either season.

I wrote about speaker wire hokum in my old audio mythology column back in the 80s.

Still, this is all getting rather cantankerous so I shall have a quick look at the notification I have and this will be my last post on this site.

This website does not have technical editors, but the moderators each have decades of industry experience and try to keep discussions on a positive, accurate path.

Good Luck

JR

 

[Matador]

This is completely wrong. A Q of 0.707 is the maximally flat Butterworth response. Get any table of filter shapes - there's one in Walt Jung's Op Amp Handbook which is online - and you'll see the Q of a 2nd Order Butterworth is 0.707.

I don't think so. According to Vance Dickason's book, the standard 4th order Butterworth response (B4, or the standard Theile alignment), has an overall response Qts of 0.383. See Table 5 (alignment #5) from Theile's original publication:

https://sdlabo.jp/archives/Loudspeakers_in_Vented_Boxes_Part_1-2.pdf

 

[Gold]

This is a really good thread. I sure hope it doesn’t spiral out of control. I like having members from the hifi world contributing.

 

[abbey road d enfer]

Quote: "Damping factor is a near meaningless figure. As Doug Self says in one of his early books, the amplifier does not have an iron grip on the loudspeaker. Not anything like."

I fully agree with that. I mentioned it only because of the assertion that "Cable may be 0.1 ohm nowadays but it didn't used to be until people started taking cable seriously!"
History shows that people took source impedance for loudspeakers seriously very early.

"Halving or doubling the damping factor will usually make virtually no difference to the response"
According to this, the cable's resistance would make even less difference.


Quote: "This is completely wrong. A Q of 0.707 is the maximally flat Butterworth response. Get any table of filter shapes - there's one in Walt Jung's Op Amp Handbook which is online - and you'll see the Q of a 2nd Order Butterworth is 0.707. "
That is true for a 2nd-order filter. A Bass-Reflex speaker is a 4th-order. Quite different alignment.

Check post#177.

"There are almost no speakers in the world with a Q of 0.3."
PHL 3411NdS: Qts=0.20
B&C 10BG76 Qts=0.22... and many others from JBL, Faital, 18Sound...


I don't want to argue with someone who prefers to escape than face the fact he is erring.
I'm just putting things on the table.

 

[Newmarket]

Determination of "best" is left to the old method of polling.
That's what Apple have done with their earbuds and headphones surveys. The conclusion is that the poll results influence the design of headphones and earbuds, which in turn influences the polls and that feeds a vicious circle. One of teh results is that a notable proportion (40% IIRC) of the youth are going to be hearing impaired.

Can you expand on what you mean here wrt Apple ? Interested in the hearing impairment implications. Thanks in advance.