Estimating channel imbalance of headphone cable
post-15185897

#### bigshot

Numbers aren't important. The sound you can hear is. Audiophiles waste tons of energy worrying about numbers, and they don't even know what the numbers represent. While they are chasing down numerical rabbit holes they ignore the big things they can actually hear. It's silly and it just goes to show how little audiophiles know about audio.

Last edited:
post-15186261

#### bobbooo

1. In which case you'd be measuring the headphones (transducers) not the cable. "But I think you knew that."

2. I'm not sure why you'd want to measure the effect of broken/faulty cables and not just buy properly functioning ones.
2a. If the left or right signal wire is faulty/broken and the other isn't, wouldn't the worst case scenario be an imbalance equal to whatever you set your output level to, say 70dB? EG. 70dB from the functioning wire and 0dB from the broken wire? With a normal, functioning cable (of the appropriate gauge for the job), I've never been able to measure a difference between the two signal wires, hence why I said "probably" (guessed).
2b. Oh the irony! Taking your conditions of 1dB threshold of audibility and a faulty cable (one of the two signal wires/connections were faulty) which resulted in a channel imbalance of 0.03dB, that would be roughly 3,400 times below audibility, not 30 times (the dB scale is logarithmic, not linear), so your calculation "is off by about 2 orders of magnitude. It's always best to check these things to be sure ..."!

Obviously though, regardless of whether it's 30 times, 3,400 times or a million times below audibility, they're all well beyond the point of being inaudible and therefore the audible difference is 0dB.

G
1. If you want to be really pedantic (which you clearly do), I wouldn't be measuring the headphone transducers, I'd be measuring the displacement of the microphone diaphragm caused by the longitudinal displacement of air molecules between it and the headphone transducer. But I think we both knew that

2. I answered this above in reply to castleofargh – curiosity, and I'm looking at ordering cables that have the occasional bad unit so was wondering if this would even have an audible effect.

2.b I’m well aware that the dB scale is logarithmic. I’m not sure where you got your 3400 figure from though. A value of 0.03dB is 33.3 times lower than a value of 1dB. But admittedly that’s not really a useful way to talk about dB differences. The relative (rather than multiplicative) difference is more useful, which is 0.03dB - 1dB = -0.97dB. This translates to 0.03dB being 89% of the sound pressure of 1dB (see the ‘Level change Δ Lp => Ratio y for sound pressure’ converter on this page: http://www.sengpielaudio.com/calculator-levelchange.htm).

Let's leave the pedantry there, because I answered my original question a while ago, and the conclusion is, in some cases, yes, resistance imbalance in real, poorly soldered (but not broken) cables, e.g. the worst ones in the above list, can cause a just audible channel imbalance, but for most cases it won't.

post-15186278

#### bobbooo

Numbers aren't important. The sound you can hear is. Audiophiles waste tons of energy worrying about numbers, and they don't even know what the numbers represent. While they are chasing down numerical rabbit holes they ignore the big things they can actually hear. It's silly and it just goes to show how little audiophiles know about audio.
This is the Sound Science section of the forum, so numbers are important, either via quantitative theoretical estimates or direct empirical measurements. That's how science works. There are plenty of other sections of this forum devoted to purely subjective arguments. Of course, the sound you hear is the most important thing in this hobby, but the fact is there are innumerable subconscious biases involved in listening to audio, so only blind testing can be trusted to provide true scientific conclusions about subjective impressions and the capabilities of the human ear.

Last edited:
post-15186470

#### bigshot

Numbers don't necessarily mean sound, and this is Sound Science, not just science. If you can't hear it, it isn't sound. It's just a measurement. Things you can't hear won't make your stereo system sound better. Applied science, not abstract science.

The thresholds of human perception have been tested and quantified. But most audiophiles have no clue where that line lies. They keep adding "a little bit more just to be safe" over and over until they're miles over the border of inaudible. It's good to do as much research into the limits of human perception as it is into digital technology. They're both just as important.

Last edited:
post-15186505

#### bobbooo

Numbers don't necessarily mean sound, and this is Sound Science, not just science. If you can't hear it, it isn't sound. It's just a measurement. Things you can't hear won't make your stereo system sound better. Applied science, not abstract science.

The thresholds of human perception have been tested and quantified. But most audiophiles have no clue where that line lies. They keep adding "a little bit more just to be safe" over and over until they're miles over the border of inaudible. It's good to do as much research into the limits of human perception as it is into digital technology. They're both just as important.
I totally agree. That's why you'll notice in my posts, I set the threshold of audibility difference to a lenient 1dB (a number!), which is consistent with the upper bounds in the literature and various studies on human hearing.

post-15186513

#### bigshot

1dB is worst case scenario with tones. It can be considerably higher in some frequency ranges and under music rather than tones.

post-15186551

#### bobbooo

1dB is worst case scenario with tones. It can be considerably higher in some frequency ranges and under music rather than tones.
I think it's a reasonable middle-ground looking at the literature, which ranges from a high of 3 to a low of just 0.25dB:

(JND=Just Noticeable Difference)

Study Authors; Year Published; Min. Detectable Fluctuation:
Reisz; 1928; ~1 dB
Dimmick & Olson; 1941; JND = 1.5 dB to 3 dB
Atal, et. al.; 1962; ~ 1 dB
Jestaedt, et. al.; 1977; JND @ 80 dB = 0.5 dB, JND @ 5 dB = 1.5 dB
Toole and Olive; 1988; 0.25 dB for a 5kHz resonance, Q = 1

Last edited:
post-15186605

#### bigshot

Have you tried it yourself? If you have, you know that .25dB is a pretty far stretch. You'd have to set up a really atypical situation to discern that. With commercially recorded music on home audio equipment at normal listening levels, it's right around 3dB. We might be a little more sensitive to differences right in the sweet spot of human hearing between 2kHz and 5kHz, but even there under normal music listening conditions, I doubt it would be easy at all to hear 1dB. 1dB is a good general ballpark figure for tones, but who listens to tones?

It's good to establish the conditions you're talking about. For most of us, that's listening to commercially recorded music at normal listening levels in the home using home audio equipment. Bleeding edge thresholds subject to a whole bunch of exceptions are interesting, but not particularly useful. Some stuff just doesn't matter when you put it in context. The real world truth is closer to the higher end of your range, not the lower.

Last edited:
post-15186969

#### bobbooo

Have you tried it yourself? If you have, you know that .25dB is a pretty far stretch. You'd have to set up a really atypical situation to discern that. With commercially recorded music on home audio equipment at normal listening levels, it's right around 3dB. We might be a little more sensitive to differences right in the sweet spot of human hearing between 2kHz and 5kHz, but even there under normal music listening conditions, I doubt it would be easy at all to hear 1dB. 1dB is a good general ballpark figure for tones, but who listens to tones?

It's good to establish the conditions you're talking about. For most of us, that's listening to commercially recorded music at normal listening levels in the home using home audio equipment. Bleeding edge thresholds subject to a whole bunch of exceptions are interesting, but not particularly useful. Some stuff just doesn't matter when you put it in context. The real world truth is closer to the higher end of your range, not the lower.
I couldn't find a proper ABX JND test using actual music, do you have a link to one? I did try the pure tone test at Audiocheck though, over here: https://www.audiocheck.net/blindtests_level.php?lvl=0.1

I passed all the tests down to 0.2dB, and although I failed the 0.1dB test, I could reliably distinguish the '0.1dB Up' shifting from the others, but I couldn't distinguish between the '0.1dB Down' and flat tones. All this, and I have tinnitus! I actually don't think test tones are irrelevant to real-world listening - my music taste is very varied, ranging from sparse classical and ambient, to alternative rock, jazz, fusion, and electronic music, some of which consists of nearly pure sinusoidal tones in passages, so not too dissimilar to test tones. If I can distinguish 0.2dB with test tones, I suspect I can distinguish 1dB with music.

(Aside: Surprisingly, I also pass the absolute phase test on Audiocheck every time as well: https://www.audiocheck.net/blindtests_abspolarity.php)

Last edited:
post-15186984

#### bigshot

Just do some tests with music. It's really easy. And realize that your ears adjust for minute changes. A direct A/B isn't the same as listening to music normally. Even if the difference is small, it doesn't matter.

I don't understand why people take pride in exceptional hearing. It isn't necessarily a good thing, and investing your ego into your ears is a recipe for bias. Focus on the broad strokes first. There are things that matter a lot more than hearing .1 dB differences.

Last edited:
post-15187109

#### gregorio

1. If you want to be really pedantic (which you clearly do), I wouldn't be measuring the headphone transducers, I'd be measuring the displacement of the microphone diaphragm caused by the longitudinal displacement of air molecules between it and the headphone transducer. [1b] But I think we both knew that
2. I answered this above in reply to castleofargh – curiosity, and I'm looking at ordering cables that have the occasional bad unit so was wondering if this would even have an audible effect.
2.b [1] I’m well aware that the dB scale is logarithmic.
[2b2] A value of 0.03dB is 33.3 times lower than a value of 1dB.
[2b3] The relative (rather than multiplicative) difference is more useful, which is 0.03dB - 1dB = -0.97dB. I’m not sure where you got your 3400 figure from though. This translates to 0.03dB being 89% of the sound pressure of 1dB (see the ‘Level change Δ Lp => Ratio y for sound pressure’ converter on this page: http://www.sengpielaudio.com/calculator-levelchange.htm).
[3] Let's leave the pedantry there, because I answered my original question a while ago, and the conclusion is, in some cases, yes, resistance imbalance in real, poorly soldered (but not broken) cables, e.g. the worst ones in the above list, can cause a just audible channel imbalance, but for most cases it won't.
1. Exactly, you'd be measuring the sound pressure waves created by the transducer, not the cable. 1b. Apparently not.

2. Surely that depends on how "bad" the unit. You mentioned a soldering fault, which obviously can be anything from negligible to the joint/connection being completely broken. Therefore:
2a. OK, you've defined a particular case (with 0.03dB signal loss) but that's effectively arbitrary.
2b1. But you're calculations ignore this fact and treat dB as linear instead of logarithmic.
2b2. No, it's not! It would be if the dB scale were linear instead of logarithmic. For example, 33.3 times lower than 100dB is NOT 3dB, it's about 69dB!
2b3. No, that's the inverse of the difference! 0.03dB is NOT 89% of the SPL of 1dB, 0.97dB is, but the difference isn't 0.97dB, it's 0.03dB! Given your scenario the question would effectively be: Can you hear the difference between 1dB and 0.97dB (IE. A 0.03dB difference), it would NOT be can you hear a 0.97dB difference! Expressed as a factor rather than a (logarithmic) dB value, the ratio represented by 0.03dB is 1.00346, which is where the "3,400 times" comes from!

3. "Let's leave the pedantry" of a completely incorrect calculation and conclusion ... really? Your conditions (a 0.03dB imbalance due to a faulty solder joint) would NOT be just audible, it would be about 3,400 times below audibility!!

G

post-15187196

#### bobbooo

1. Exactly, you'd be measuring the sound pressure waves created by the transducer, not the cable. 1b. Apparently not.

2. Surely that depends on how "bad" the unit. You mentioned a soldering fault, which obviously can be anything from negligible to the joint/connection being completely broken. Therefore:
2a. OK, you've defined a particular case (with 0.03dB signal loss) but that's effectively arbitrary.
2b1. But you're calculations ignore this fact and treat dB as linear instead of logarithmic.
2b2. No, it's not! It would be if the dB scale were linear instead of logarithmic. For example, 33.3 times lower than 100dB is NOT 3dB, it's about 69dB!
2b3. No, that's the inverse of the difference! 0.03dB is NOT 89% of the SPL of 1dB, 0.97dB is, but the difference isn't 0.97dB, it's 0.03dB! Given your scenario the question would effectively be: Can you hear the difference between 1dB and 0.97dB (IE. A 0.03dB difference), it would NOT be can you hear a 0.97dB difference! Expressed as a factor rather than a (logarithmic) dB value, the ratio represented by 0.03dB is 1.00346, which is where the "3,400 times" comes from!

3. "Let's leave the pedantry" of a completely incorrect calculation and conclusion ... really? Your conditions (a 0.03dB imbalance due to a faulty solder joint) would NOT be just audible, it would be about 3,400 times below audibility!!

G
You've misread my previous posts. If you re-read them you'll see I said "From my calculations, worst case scenario it looks like it might [be audible], at 1dB imbalance. But for the cables I'm looking at it's a max imbalance of around 0.03dB with my setup." I'm claiming 1dB channel difference could be audible, not 0.03dB. That's my conclusion.

'''How many times" lower 0.03dB is than 1dB is really not important here, as if it's inaudible, it's inaudible. But I'll play along. How do you go from a ratio of 1.00346 to "3400 times"?

Last edited:
post-15187233

#### bobbooo

Just do some tests with music. It's really easy. And realize that your ears adjust for minute changes. A direct A/B isn't the same as listening to music normally. Even if the difference is small, it doesn't matter.

I don't understand why people take pride in exceptional hearing. It isn't necessarily a good thing, and investing your ego into your ears is a recipe for bias. Focus on the broad strokes first. There are things that matter a lot more than hearing .1 dB differences.
A blind ABX test is really the only scientifically valid test in psychoacoustics, in order to eliminate any unconscious biases present. And it's not that easy to set up, especially for real music. But I'll see what I can do with somes music files, Audacity and Foobar (I think the latter has an ABX tool).

Oh and I'm not "taking pride" in my hearing at all, I was just honestly surprised by my results considering I have quite bad tinnitus. I disagree with your statement that exceptional hearing isn't necessarily a good thing though. Of course it is. No-one would claim that about vision, and hearing should be no different.

Last edited:
post-15188320

#### bigshot

It's not hard at all to do a simple controlled test. It's the best way to understand the relative importance of things. On paper, something may seem important, but when you hear it on your system with your music and your ears, it may not mean anything at all.

When I was a kid, I must have had a very good ear for ultra high frequencies. I would go to the Sears store and get sick from the high frequency squeals coming from the ballasts in the overhead fluorescent lights. I've always been sensitive to sharp dynamics in the sweet spot of hearing too. I had to wear cotton in my ears at recording sessions at one point. There are people who can't focus past surface noise to enjoy the music on historical recordings. People with perfect pitch can detect problems in musical performances that other people aren't bothered by. The kind of hearing that counts is the part of your hearing that appreciates music. That doesn't require exceptional ears at all. Normal human ears do the job perfectly.

post-15188429