Capacitor Sound

[tedmich]

most commonly these hifi guys mock up a specific circuit that allows them to differentiate between capacitor types, they NEVER actually do double blind tests which show they can differentiate between cap types in their specific application; reproducing audio.

Check the ABX Comparator discussion:
The Truth Should Out | Stereophile.com
The Highs & Lows of Double-Blind Testing | Stereophile.com

Audio, Meet Science | Stereophile.com

a favorite:
Do Coat Hangers Sound As Good Monster Cables? - The Consumerist

[Rhodesplyr]

J.M.,

I shouldn't have gotten involved in this discussion since I know that it's one that gets people very worked up.

However, since I used to be a professional academic, I might ask this question: Has Bateman's work been peer-reviewed, and have his results been independently replicated and verified? If we're talking rigorous science here, those are standard procedures. You ask if I read Bateman's actual measurements. No, I didn't read them, and, for that matter, neither did you. I read Bateman's *written account* of his measurements, as did you. In fact, I'm sure I read this article some time ago. It's certainly not the first time I've run across it.

I'm not disputing Bateman's results. I'm simply asking if they have been verified. Since the miracle of cold fusion and the danger of vaccines that cause autism, I've grown slightly more skeptical of the things that pass for knowledge :-)

[tedmich]

audiophiles often maintain that science cannot deal with their "golden ears"
Audiophiles and the limitations of human hearing « The Phineas Gage Fan Club

[Steve Conner]

See, this is why I kept quiet about those articles.

There is some truth to the "science can't handle Mah ears" kind of argument. But it is more like "audio manufacturers peddle pure crap and write utter bullshit specs, but a good pair of ears can expose the truth if you're lucky."

The ear is more sensitive to high order harmonics than you would expect, though scientists acknowledge this. And solid-state hi-fi amps produce much more distortion than the spec sheet figure implies, because that figure is a marketing one.

So it is possible for two hi-fi amps to sound slightly different even though the manufacturers claim they both have 0.001% distortion, which would of course be inaudible if it were true. But if you measure them yourself, you'll see the distortion is actually a lot worse than claimed and does correlate with the subjective sound impression.

I can believe that different capacitors sound different in crossovers too.

Most problems in audio are really gross. I once had an EQ box that made an audible and unpleasant change to the sound. Turns out it was made with Class 2 ceramics in the low frequency section, introducing 0.25% THD. And the boost/cut knobs were installed squint so that the response wasn't flat when set to "0".

All of this is of course hi-fi and nothing to do with guitar amps. I once tried measuring THD of a dirty channel and got a result of 10,000% from a badly confused test set.

[J M Fahey]

Dear rhodesplayer.
Please don´t consider this personal, at all.
I didn´t ask *you* specifically but my question was general, addressed to all interested.
You are right, we are reading Bateman´s article, not an independent verification.
Yet it *looks* like he made a great effort in being neutral , the oscillator design looks very polished , the spectrum analyzer images look like screen captures.
Might be Photoshop or Corel Draw, of course

On another point, I *do* believe that the human senses sometimes can perceive something that test instruments can not.
Not blaming the very good lab equipment we can build today, but "bad design of the experiment" which might be measuring with incredible precision a parameter which really is not the relevant one.

However, to deal with the "human Lab" results (our brains and ears), the only *real* accuracy comes from double blind tests.
Why?
Because we are "calibrating" it against itself.
Meaning "I can hear something when I use xxxx" but I don´t really know whether xxxx was actually used, or was it yyyy?
And the guy who´s checking me doesn´t either (of course a "control" third guy does) so he can´t cue me even by the tone of his voice or the way he looks at me.
Under those circunstances, I can trust the "imperfect Human Lab" very much.
Funny that so many feel outraged when such a simple and honest testing is suggested.

[Steve Conner]

Would you actually use ABX testing while designing a musical instrument amp or effect?

I'm trying to design a simple ABX system just now. It would be easy to write a simple program like WinABX that would switch a bunch of relays and analyse the results of the trials. Indeed you could use WinABX itself if you made the relay box take tones of different frequencies to select A or B.

The problem as I see it, at least for hi-fi power amps, is the switching. I've seen two occasions where relay contacts make more distortion than my prototype power amps. They work great out of the box, but maybe they pick up a tarnish from air pollution.

[J M Fahey]

Dear Steve.
Don´t have the faintest idea what ABX might mean, but yes, I was thinking that a Computer system (very simple, even a humble PIC would do) that switched whatever I´m trying to test in and out, alternating with what I´m comparing it to, following a random or pseudo random sequence of which *I* would not be aware of, but which the system would duly register, would be trusty.
Maybe, as you say, the relay contacts could introduce more distortion than your amp's but there´s a zillion other things to compare:
different capacitor dielectrics, brands or colors, different tube brands, PTP vs PCB, different speakers, you name it.
The system would be very simple: beep or blink a Led, indicating a change (it might have changed or not, that´s the point), and I´d tick a box on a sheet of paper.
Better/worse , smooth/raspy , whatever.
Even "reduced/expanded soundstage"
Afterwards, let´s compare the hand ticked perceived results with a computer printed sheet of what actually happened.
It provides only the "accounting", the "measuring instrument" is myself.

And, as you know, 50% accuracy, which is a typical blind test end result, is the same as flipping a coin

[Steve Conner]

ABX testing is essentially what you described. You set up two different sounds, A and B. Then you do a number of trials where the computer selects one or other of them at random (that's X) and you have to say whether you think it is A or B. You can flip between A, B and X and replay them as often as you like before answering.

As you do more trials, the computer displays your percentage success, and the confidence level that this is not due to chance, which increases with the number of trials.

This just proves that you can distinguish A from B, it doesn't help you decide which one you prefer. I like to think of that as a separate test to be done once you've shown that there is a difference.

WinABX does this for you using two audio files. I've had some interesting times with it. Once I was comparing two files from mic preamps, where someone on the home recording forum was able to tell them apart very well. He described one of them as sounding bigger and more spacious, and everyone else laughed. After some practice I was able to tell them apart too, and FFT analysis showed that the "bigger" sounding one had lots of low-frequency noise that gave a deep, rumbly quality to the noise floor. It had nothing to do with the music that was being recorded!

I liked your comment about a great experiment that fails because it measures the wrong thing. I believe that every audible difference correlates to something measurable, if you only knew what to measure.

[teemuk]

Would you actually use ABX testing while designing a musical instrument amp or effect?

I don't see why designing something that pleases ear and using ABX blind listening tests to find out whether certain stuff is actually even perceivable would be mutually exclusive techniques.

Yes, things that generate coloured sound are ultimately designed to please the designer(s) ears; usually with good faith that a considerable bunch of other people (read: potential customers) will share the same taste with the designer. There's no denying of that.

I still don't see why such technique would exclude, for example, A/B listening tests that determine whether you really need a $5k power cord, $$$ NOS tubes or brand capacitors instead of just inexpensive, generic ones, or if the 1% tolerance of certain component is essential for achieving The Tone in oppose to using 5% or even 20% tolerance. And so on.

It's not like things have to be designed either 100% by ear or 100% by looking at measuring equipment and statistical results of blind listening tests. Why does this topic always bring out such polarized, black&white view of the matter?

[Steve Conner]

Yes, I totally agree. This is what the big names like Harman do, and it's why I'm interested in building an ABX setup that works with real hardware as well as digital audio files.

I think the reason for the polarisation is that the objectivists are all boring and would never dream of tuning anything by ear, and the subjectivists all run boutique power cord stores. ;-)

I'll leave you with this. Everyone knows vinyl is superior to digital, but one Stereophile reviewer uses the $1k Alesis Masterlink digital recorder to record the output of $100k turntables for comparative listening tests. No matter how you take that, it's a level of cognitive dissonance that would cause "lesser" minds than Stereophile's to implode.

[tedmich]

ecologically horrid Hg relays are still available and have very clean contact responses, and surely a randomization algorithm with multiple relays in the path could average out relay distortion for each sample? Of course getting enough different pairs of ears to listen to your ABX rig is also important, unless you just want to discover your personal truth.

[J M Fahey]

one Stereophile reviewer uses the $1k Alesis Masterlink digital recorder to record the output of $100k turntables for comparative listening tests. No matter how you take that, it's a level of cognitive dissonance that would cause "lesser" minds than Stereophile's to implode.

Maybeeeeeeeeeeee ........................... because Stereophile´s brains have *already* imploded?

[Wombaticus]

J.M.,

I shouldn't have gotten involved in this discussion since I know that it's one that gets people very worked up.

However, since I used to be a professional academic, I might ask this question: Has Bateman's work been peer-reviewed, and have his results been independently replicated and verified? If we're talking rigorous science here, those are standard procedures. You ask if I read Bateman's actual measurements. No, I didn't read them, and, for that matter, neither did you. I read Bateman's *written account* of his measurements, as did you. In fact, I'm sure I read this article some time ago. It's certainly not the first time I've run across it.

I'm not disputing Bateman's results. I'm simply asking if they have been verified. Since the miracle of cold fusion and the danger of vaccines that cause autism, I've grown slightly more skeptical of the things that pass for knowledge :-)

As I pointed out earlier in the thread, Douglas Self has replicated and largely verified many of Bateman's key results. And while Self himself appears to be alarmed by some of the figures (he's evidently a "chasing zero" type of objectivist), it's pretty clear that for MI applications even worst case cap distortion would be inconsequential. For those that care anyway, the punchline was that polypropylene caps were virtually distortion free.

[Rhodesplyr]

As I pointed out earlier in the thread, Douglas Self has replicated and largely verified many of Bateman's key results. And while Self himself appears to be alarmed by some of the figures (he's evidently a "chasing zero" type of objectivist), it's pretty clear that for MI applications even worst case cap distortion would be inconsequential. For those that care anyway, the punchline was that polypropylene caps were virtually distortion free.

Thanks for pointing that out. Self's book looks like an interesting read. One thing I noticed that he did mention, however, was that he found some capacitors he tested that had problems as an apparent result of poor build quality, not a result of the nominal dielectric material.

Here's a case I encountered. I was rebuilding my own Leslie 122 amp, and I was out of 0.015uF capacitors, which you find right at the input. I was out of these and was getting some parts from a Hammond/Leslie parts dealer, so I asked him if he could throw in a couple, which he did. They were very small, little blue squares. In any case, I fired the amp up, and though it worked, it didn't sound right. The best word I can think of to describe it is "anemic." Since all other parts were standard ones I'd used before, I ordered some SBE 716P units--same value, same voltage. I soldered these in, and all was well. The small blue caps tested at the correct value, but something was wrong with them. Poor lead attachment? Substandard dielectric material? I don't know.

I've done some investigation into the history of capacitor types, and, based on what I read, there are many technical challenges to making them consistently, from the production and quality control of the dielectric material to maintaining proper winding tension for wound types, cleanliness of the manufacturing facility, etc... In other words, there are lots of places for things to go wrong that have nothing to do with more general issues of polystyrene vs. ceramic vs. polypropylene, though, in the absence of information to the contrary, these issues might get *attributed* to the dielectric since that's what most people know about.

[Diablo]

OH no! Has someone destroyed the blue Astron cap market in one fell swoop?