Golden Ears Fail Again

[FLZapped]

John,

You're falling into the wrong trap.

Please show me anything beyond phase response, and group delay and the other well known electrical parameters that define a cable can materially affect the "soundstage."

Those are very measureable properties, are they not?

Of course I realize that soundstage is a complex variation beginning with the acoustic space where the recoding originated and ends with the acoustic space where the listener receives the results of the recording.

However, this conversation revoloves around the effects of cables. And the parameters that would affect "soundstage" in any way are indeed measurable. Whether or not there is a known and direct corallary, is another story. It STILL doesn't exist outside resistance, capacitance, and inductance and the effects these parameters impart on the signal. Does it?

Therefore, if you had two similar cables that had different group delay measurements and were able to verify a shift in "soundstage" - it wouldn't be hard to postulate that the group delay differences had something to do with it, now would it?

Next time, I'd appreciate it if you would be factual in your comments without being condescending.

-Bruce

[musicoverall]

What you fail to understand is that he did not purchase them thinking they were making his system sound better. Maybe he just thought they were sexy. Of course, he coulda won them on a bet, too......maybe you should email him and ask.


-Bruce

How do you know he didn't purchase them thinking they were making his system sound better? I didn't notice that he mentioned anything about the time of purchase.

Maybe the cow jumped over the moon, too!

[jneutron]

Next time, I'd appreciate it if you would be factual in your comments without being condescending.

Condescending? You have no idea what condescending is...actually, neither do I...

Sorry Bruce, it was not meant to be cond...that word...It was supposed to be posed as a logical sequence of things necessary to produce a soundstage..guess thinking about my response to clark johnson tailored my phrasing...:-)...however, it was factual..

You're falling into the wrong trap.
Please show me anything beyond phase response, and group delay and the other well known electrical parameters that define a cable will materially affect the "soundstage."

I answer that question with a question:hope I don't come across conde...you know..

Describe for me how to measure the time relationship between the currents and current slew rates for two audio, low impedance, channels, while the amplifiers are running in all four quadrants, with the bulk of the power drive having no correlation between channels... with the signal of interest buried within the power signal. And, measure it across the entire frequency band, to the 1 uSec level.

If you tell me you measured voltage, then you didn't measure the current or current slew. Voice coil drivers accelerate based on current, not voltage..hanging a distributed inductance/capacitance/resistance between the voltage controlled node and the reactive low impedance node, doesn't guarantee that what you see at the node is what is at the amp terminal..indeed, just attempting to measure the voltages at both points introduces measurement errors..

So far, the best measuring instrument I can think of is between our ears..

Those are both very measureable properties, are they not?

Using sine wave excitation, it is rather easy to measure amp and phase..but if a cos theta signal is added to one of the frequencies in a complex signal, do you think it is easily measured? FFT's come up zip in that respect.

Of course I realize that soundstage is a complex variation beginning with the acoustic space where the recoding originated and ends with the acoustic space where the listener receives the results of the recording.
However, this conversation revoloves around the effects of cables. And the parameters that would affect "soundstage" in any way are indeed measurable. Whether or not there is a known and direct corallary, is another story. It STILL doesn't exist outside resistance, capacitance, and inductance and the effects these parameters impart on the signal. Does it?

Last part first..

I have never, ever, claimed that RLC is not the whole story...I think maybe you've confused me with jon?? It ain't magic..it is still a question of how RLC can affect soundstage. I'm not losin it...not yet, anyhoo..

In my research, I suddenly hit a brick wall..to what level do I have to go to in order to measure how RLC can affect soundstage?

The questions I posed you (perhaps not in a very nice way...again, sorry) is exactly what needs to be answered, before testing can confirm (or deny) the existance of an effect between cables and soundstage..

Therefore, if you had two similar cables that had different group delay measurements and were able to verify a shift in "soundstage" - it wouldn't be hard to postulate that the group delay defferences had something to do with it, now would it?

That'd be a good postulate..although group delay, I'm not convinced is the correct way to look at it..especially for a reactive load, and a transmission line about 6 orders of magnitude below wavelength.

( of course, you can't forget about the grain boundary issue, the dielectric involvement, the motor-generator effect, the piezo effect, strand jumping, skin effect, the jitter effect (hey waittaminute, is that a cable thingy, or a CD thingy...It's so difficult to keep this garbage straight...the world of "high end audio" is so full of foolish guru wannabe's and charlatans, no wonder a quantum dot CD fixer selling for 40 bucks sells..

Cheers, John

Oh, forgot to add..

I have been discussing with an audiophile guy, testing exactly what you said...determining correlation between RLC, and what is heard..

Since I can design and build speaker cables while changing only one parameter at a time, for example, making two sets, one of 4 guage, one of 18, both 7.6 nH per foot/135 pf per foot, It will be easy to see what preferences may arise..it'll take a while, but it'll be fun.

[Monstrous Mike]

You are absolutely correct...and, a master of understatement..

Hi Mike..happy new year..

Cheers, John

Thanks, same to you.

[musicoverall]

I figure about a year on each task..

Your motivation is inspiring! Carry on, sir!

[FLZapped]

Describe for me how to measure the time relationship between the currents and current slew rates for two audio, low impedance, channels, while the amplifiers are running in all four quadrants, with the bulk of the power drive having no correlation between channels... with the signal of interest buried within the power signal. And, measure it across the entire frequency band, to the 1 uSec level.

Why doest thou digress off topic thus?

I don't care. (cue etherial music) Cables John, not amplifiers. Passive, not active. Topic at hand, okay?(fade music)

I'm sure they're important, but not for this discussion....at least not yet.....

I have never, ever, claimed that RLC is not the whole story...

Never said you did. What I said was, or at least should have said, is that no one has been able to show that it is anything other than these parameters. Unless it's JPS Lab's digital resistance. ;~} (Maybe tht's the answer, ya think?)

I think maybe you've confused me with jon?? It ain't magic..it is still a question of how RLC can affect soundstage. I'm not losin it...not yet, anyhoo..

No confusion, well, not identity wise. And yes, finally we are on the same page, maybe it's the same floor. I think I lost it a while back, seriously, I have so many things lately I am expending enormous amounts of emotional energy on, I can't concentrate enough these days to get into heavy discussions. My thinking processes are just too crowded.

In my research, I suddenly hit a brick wall..to what level do I have to go to in order to measure how RLC can affect soundstage?

Welp, maybe the answer lies somewhere in specification for the surround sound systems in movie theaters, or that 30 speaker monster sphere that NASA has created. (I think it was 30, can't find the bloody link now)

The questions I posed you (perhaps not in a very nice way...again, sorry) is exactly what needs to be answered, before testing can confirm (or deny) the existance of an effect between cables and soundstage..

Well, maybe it should be simplified to just finding out the perception threshold in measured differences first....especially group delay.(I agree with Steve Eddy on this one)

That'd be a good postulate..although group delay, I'm not convinced is the correct way to look at it..especially for a reactive load, and a transmission line about 6 orders of magnitude below wavelength.

6 orders of magnitude? I think there is part of a sentance missing here......*lost*

Perhaps. Certainly it would be a bit easier in the interconnects than the speakers cables since there is less reactive variation of the loads. However, the starting point is always a purely resistive load....well, as close as you can get to one, anyway and then us a load simulator and watch what happens as you pull the load around the smith chart....oh wait. that's rf....nevermind....or maybe.....

( of course, you can't forget about the grain boundary issue, the dielectric involvement, the motor-generator effect, the piezo effect, strand jumping, skin effect, the jitter effect (hey waittaminute, is that a cable thingy, or a CD thingy...It's so difficult to keep this garbage straight...the world of "high end audio" is so full of foolish guru wannabe's and charlatans, no wonder a quantum dot CD fixer selling for 40 bucks sells..

JOHN! hahahahhahahhaa.....

Cheers, John

Oh, forgot to add..

I have been discussing with an audiophile guy, testing exactly what you said...determining correlation between RLC, and what is heard..

Since I can design and build speaker cables while changing only one parameter at a time, for example, making two sets, one of 4 guage, one of 18, both 7.6 nH per foot/135 pf per foot, It will be easy to see what preferences may arise..it'll take a while, but it'll be fun.

Well, have fun. Personally, I think I'd start with the less reactive situation of interconnects and then graduate to speaker cables....but I know how you loath the easy path. ;~}

Need a bruised nailbed?

-Bruce

[FLZapped]

How do you know he didn't purchase them thinking they were making his system sound better? I didn't notice that he mentioned anything about the time of purchase.

Maybe the cow jumped over the moon, too!

Probably because I've known eyespy a hell of a lot longer than you have...which for you is not at all.

Sorry if you missed the last showing of the cow jumping over the moon thingy, it's gonna be a while, it's all sold out for the next millenium. The part for the dish and spoon is open, we can't figure out where they ran off to.

-Bruce

[RobotCzar]

So... if I were to conduct some blind tests, how would I go about it? The long term listening tests outlined by (as I recall) Magictooth will, I'm told, do nothing to sway the scientific community. Instead, I need to conduct tests with about 30 second intervals. True? Let's further go out on a speculative limb here and say I pass with statisical significance. What happens then?

Nobody expects you to set up totally valid scientific experiment, or to run valid statistical tests. But, your goal is the same as a valid scientific study: to remove bias and other factors that could account for the differences you are hearing. This basically means that you should attempt to make the two alternatives you are trying to distinguish as similar as possible. For example, if testing to hear difference in amps, have the rest of the system be the same (e.g., don't use differnt sources or speakers for the things being compared). The factors most often overlooked are:

1) You must make sure whatever you are comparing are at the same volume level. this is trickier that you think as very small differences in volume can be detected by humans and the louder is often heard as better. You can use a volt meter connected to electronics or a sound pressure level meter. You probably don't have either so you can't run even a simply valid test. Your only alternative is to play a 1 KHz sine wave (spring for a test disc) and earball the levels until you can't tell a difference (test yourself).

2) You must blind yourself (not with a blindfold, you must not know which is which when you are making a judgement. There is not a shadow of a doubt that expectation does and will affect what you hear. It is not that you are intentionally bias, your perceptions actually change so that you precieve the thing you expect to sound better as actually sounding better. This is why "just listening" is really bad adivce.

You absolutely do not have to listen for any length of time. Listen and switch back and forth as often and as quiclkly as you like. Then make a judgement. Choose your own stereo with recordings you know really well and segments you think might show a difference (or which you think do show a difference).

Check your final score. (The expectation is that you will get half correct.) Even if you got them all right it proves nothing as you may have done so by luck. But, if you can repeatedly do so, let us know. Those running a real statistical analysis would accept that you can hear a difference if you score better than would be expected by chance in one out of twenty cases of random guessing. You can choose whatever level seems reasonable to you, but I wonder what that level isn't 100% if there are "huge, obvious" differences.

An ABX devices is a switch that automatically (and secretly) chooses input A or input B and play on the X position. You can move the switch between A, B and X as often and and as slowly as you want to. Is X A or B? Choose. I think you can by these things, but having somebody else do the switch will single blind the experiment and may be okay for your personal peferences. A double blind insists that the person doing the switching also not know which is which,

[musicoverall]

Probably because I've known eyespy a hell of a lot longer than you have...which for you is not at all.

Sorry if you missed the last showing of the cow jumping over the moon thingy, it's gonna be a while, it's all sold out for the next millenium. The part for the dish and spoon is open, we can't figure out where they ran off to.

-Bruce

Well, since you know him, ask him to get some new links for his website, will ya?

So the cow did jump over the moon? Hmmm... I guess you're right about someone buying Nordost SPM for looks, then.

[musicoverall]

Nobody expects you to set up totally valid scientific experiment, or to run valid statistical tests. But, your goal is the same as a valid scientific study: to remove bias and other factors that could account for the differences you are hearing. This basically means that you should attempt to make the two alternatives you are trying to distinguish as similar as possible. For example, if testing to hear difference in amps, have the rest of the system be the same (e.g., don't use differnt sources or speakers for the things being compared). The factors most often overlooked are:

1) You must make sure whatever you are comparing are at the same volume level. this is trickier that you think as very small differences in volume can be detected by humans and the louder is often heard as better. You can use a volt meter connected to electronics or a sound pressure level meter. You probably don't have either so you can't run even a simply valid test. Your only alternative is to play a 1 KHz sine wave (spring for a test disc) and earball the levels until you can't tell a difference (test yourself).

2) You must blind yourself (not with a blindfold, you must not know which is which when you are making a judgement. There is not a shadow of a doubt that expectation does and will affect what you hear. It is not that you are intentionally bias, your perceptions actually change so that you precieve the thing you expect to sound better as actually sounding better. This is why "just listening" is really bad adivce.

You absolutely do not have to listen for any length of time. Listen and switch back and forth as often and as quiclkly as you like. Then make a judgement. Choose your own stereo with recordings you know really well and segments you think might show a difference (or which you think do show a difference).

Check your final score. (The expectation is that you will get half correct.) Even if you got them all right it proves nothing as you may have done so by luck. But, if you can repeatedly do so, let us know. Those running a real statistical analysis would accept that you can hear a difference if you score better than would be expected by chance in one out of twenty cases of random guessing. You can choose whatever level seems reasonable to you, but I wonder what that level isn't 100% if there are "huge, obvious" differences.

An ABX devices is a switch that automatically (and secretly) chooses input A or input B and play on the X position. You can move the switch between A, B and X as often and and as slowly as you want to. Is X A or B? Choose. I think you can by these things, but having somebody else do the switch will single blind the experiment and may be okay for your personal peferences. A double blind insists that the person doing the switching also not know which is which,

Thanks. I'm almost positive a friend of mine owns an SPL meter.

BTW, who ever said cable differences are "huge" and "obvious"? I didn't.

[jneutron]

Why doest thou digress off topic thus?
I don't care. (cue etherial music) Cables John, not amplifiers. Passive, not active. Topic at hand, okay?(fade music)
I'm sure they're important, but not for this discussion....at least not yet.....

The assumption that cables can treated as an independent entity is not scientific. They are part of a chain of things that transfer energy to the air outside your ears.. I am not confident enough that it is possible to separate the blocks in this fashion, as they inter-react at some level.

If the amp reacts strangely to the incremental inductance changes we are talking about (5 nH per foot up to 200 nH per foot, this reaction is faster than the normal 20 to 20K human hearing capability, meaning it isn't on the radar screen of audio testing..I must admit, finding initially that we need to discriminate ITD's to the level of 1 to 5 uSec really threw me. I had never considered that kind of current slew rate before.. Once those numbers came up, I realized rather quickly (coupla picoseconds), that standard test methods are out..I've dealt with low impedance measurement errors at high current slews in the past. If you don't consider the geometry of the setup in terms of e/m field theory, accurate measurements are doomed....(doomed, I tell ya). Using a pair of Dale 250 N style bifilar resistors, as JR trusts, is just fooling yourself..Hell, I told him that about two years ago.. I get 1.5 volt errors at 20Khz, into a 250 u-ohm resistor, if I use "standard techniques". Because the errors are cos theta errors (shifted 90 degrees relative to the actual current), seeing the actual current is not possible with standard techniques. Errors in a 1 ohm load at 4 kiloamps per microsecond was a tad bit more "challenging", but it was still possible to get accuracy to within about 1 to 5% once the mag fields were understood.

The amp? It gets it's voltage feedback directly from the output binding posts (assumption)...at high z load, there will be no current slew (I dot) error associated with measuring the voltage at those terminals.. But as load z decreases, the current and I dot error increase, but the amp AND the voltage measuring device(scope) at the output terminals, will see the exact same thing. It won't be correct, as both entities are seeing the exact same error....amp says "yada", scope says "yada", but "yada" is actually "yada-goop", goop being the error. ( I love the technical terms).

The first things to come out of my research:

1. A .1 and .01 ohm current viewing resistor capable of 1% accuracy at slew rates up to about a Kiloamp per nanosecond.
2. A load resistor geometry capable of pure resistance with the same slew spec.
3. A voltage pickup device capable of zero I dot error.
4. An air toroid I dot pickup feedthrough...this will allow direct viewing of the current slew, which is the hottest contender for ITD errors in a cable.
5. A 3 axis B dot pickup coil set, to measure changing magnetic fields around components.

I've already built revision 1 of item 2, had difficulty measuring below a nanohenry.

Item 1 is a relative no brainer, I've not yet bought the resistors.
Item 2 has gone through 2 revisions...rev 2 is the sub nanohenry unit I'm testing.
Item 3 is a triviality, so I've held off.
Item 4 is designed, but that #49 guage wire is a real Bi*ch to use.
Item 5....same Bi*chy wire in item 4.

Unless it's JPS Lab's digital resistance. ;~} (Maybe tht's the answer, ya think?)

WHAT????holy mackeral..

I think I lost it a while back, seriously, I have so many things lately I am expending enormous amounts of emotional energy on, I can't concentrate enough these days to get into heavy discussions. My thinking processes are just too crowded.

Welcome to my world...hey, didn't you have some medical problem recently? Hope all is well.

Welp, maybe the answer lies somewhere in specification for the surround sound systems in movie theaters, or that 30 speaker monster sphere that NASA has created. (I think it was 30, can't find the bloody link now)

While the research is quite good (unfortunately, not well published), the "put the sound anywhere for everyone, just ain't tenable..the dipole field cancellation technology requires a sweet spot in space, and two people cannot share it..lots of speakers, with huge dsp to move the sound around the theatre, that's cool..I had considered trying that with my 128 4 inch speakers just for kicks. But, I'm too busy. got a honeydew list the size of manhattan.

Well, maybe it should be simplified to just finding out the perception threshold in measured differences first....especially group delay.(I agree with Steve Eddy on this one)

Steve Eddy???? Where??
I'm workin on this, with a computer controlled slide for moving a single speaker, and some programming to randomize ITD and IID with some speakers, all behind a visually opaque, audibly transparent screen, front lit to hide the speakers. Intent is to use real sounds, like a cowbell or something, to develop localization relationships..I have difficulty believing the use of sines or sine amplitude modulated sines are useful for actual complex sound localization...for the researchers, it makes the problem easier, but I think it's missing the boat.

6 orders of magnitude? I think there is part of a sentance missing here......*lost*

It was a glib statement reflecting the length of an audio signal wavelength vs the length of a wire.

Certainly it would be a bit easier in the interconnects than the speakers cables since there is less reactive variation of the loads. However, the starting point is always a purely resistive load....well, as close as you can get to one, anyway and then us a load simulator and watch what happens as you pull the load around the smith chart....oh wait. that's rf....nevermind....or maybe.....

Smith charts...man, I loved using those...it was so darn easy for me to understand, while my classmates were clueless..one a dem "kodak" moments..had few, so relished every one of em...
Item 4 will be also designed to view the shield currents in the IC as a result of loop pickup of b dot due to the power cords...but that's another story.

Well, have fun. Personally, I think I'd start with the less reactive situation of interconnects and then graduate to speaker cables....but I know how you loath the easy path. ;~}

That transparent, am I?

Need a bruised nailbed?

No thanks...already got one..honeydew related..
Cheers, John

[E-Stat]

The assumption that cables can treated as an independent entity is not scientific. They are part of a chain of things that transfer energy to the air outside your ears.. I am not confident enough that it is possible to separate the blocks in this fashion, as they inter-react at some level.

Amen.

Unless it's JPS Lab's digital resistance. ;~} (Maybe tht's the answer, ya think?)

Don't mind Zapped. His imagination far exceeds his reading retention ability. It took me three posts to explain the concept of a product line to him.

rw

[theaudiohobby]

NO! You are making things up.... both are in the time domain.

I am making thing up eh , well here goes, have a read, a quote from the article goes

The frequency domain provides an alternative description of sound in which the time axis is replaced by a frequency axis. In the frequency domain, sounds are represented in a frequency by amplitude and/or phase diagram

I did read your commentary on a 'properly performing' amplifier and decided you got the point , so there is no need to heckle you further.

Happy reading

[RobotCzar]

I am making thing up eh , well here goes, have a read, a quote from the article goes

I did read your commentary on a 'properly performing' amplifier and decided you got the point , so there is no need to heckle you further.

Happy reading

Yah, but I'm not done heckling YOU. First off, can't you come up with a better reference than this? Nothing this person says is drastically wrong, and it is pretty simple. I could not find a reference to the credentials of the writer. Don't you think that is, um, kinda important?

Let me quote fromt the start of the article: "The representation of sound in the time domain is important to understand, but in some ways it is also awkward." I'll say its awkward, as sound is completely in the "time domain"--without time there is no sound, no frequency-- it IS rather difficult to talk about it in another way. But, people persist. I have asked you questions, which you refuse to answer. Why not tell me how this article contradicts what I said? YOU read it, and you will find that there is nothing in it that says that frequency and time variations in maximum or RMS amplitude are separate or in separate "domains". The term "domain" in this case is means "topic"--mere a way to organize what the author wants to say. The fact that the author uses the term "time domain" merely suggests that he isn't really an electrical enginieer or physicist. Even if he was, this label is being used as a topic, not suggesting that frequency, amplitude, and phase are not completely related. Nothing in this article suggests their would be audible differences in amplifers because of frequency and time domains.

I guess you argument runs like this... "Hey, I found an article on the Internet (from a speech college, for heavens sake---why not the physics or engineering departments?) that mentions "time domain" in regard to sound. Yay, that means all the dumb stuff said about these 'domains' is true."

Sorry, why not bite the bullet and learn something hard instead of taking the easy way out and reading technobabble? Please state what you think the article says is the difference in the frequency "domain" and time "domain", or give me your understanding if you find the article too difficult.

Next time you do a Goggle search on something, page down a little further and actually read what you find.

[theaudiohobby]

RobotCzar,

You hold an extreme and wrong opinion very much like the very individuals you heckle, just an opposite extreme, I saw other articles that dealt with the topic more rigorously but decided they were too heavy going for the purposes of this discussion, hence I put that one up, but I suppose that accepting that certain performance criteria exist in the time domain will ruin your "All components except loudspeakers sound the same" extreme mantra, dogmatic ignorance by another name, the extreme opposite of everything produces an audible difference. Here is another reference to time domain performance in digital filters and another on Class D power amplifiers.

To achieve full bandwidth operation there are several filter points that must be modified. They are the output filter, the feedback filters, the error amplifier filter and the input filter. If any of the filter points are too low there will be large phase shifts in the upper audio frequencies reducing the resolution and clarity of the highs. For this reason the frequency response of the system should be flat out to 20kHz. The mistake is often made to set the .3dB point near 20kHz resulting in good bench performance but poor quality in listening test.

The quote refers to both the phase shift (time domain) and frequency response (frequency domain) performance of the power amplifier module, i.e audible tradeoff between the frequence domain performance ( 0.3dB at 20Hz) and the time domain performance (phase shift in the upper audio frequencies) and recommends a preferred tradeoff.

You entitled to disagree with the new references but you will be wrong and like many of the dogmatic subjectivists you heckle, simply too strong-headed or egotistic to accept that your dogma is wrong, way too extreme, blind to other petinent variables, to be right.

[RobotCzar]

Dear theaudiohobby,

Here is a link I got by doing a Goggle search on "flat earth" (from a University). I had over 6 million search hits!

http://www.indiana.edu/~ensiweb/lessons/flaterth.html

The Earth is flat, the flat earthers have a point (after all, if this many people say the Earth is flat, then there must be something to it). Scientists who maintain the Earth to be sphrerical really have not proven this fact (or really that haven't proven that the Earth is NOT flat). Science has been known to be wrong many times in the past. I am kind of amazed that scientists have such a closed mind about this issue. There are clear flaws in the tests used to determine the shape of the Earth. I also can't understand why these scientists want to spoil the fun of the flat Earthers, after all they are so sincere.

Those scientists are all so snooty. And, they are biased in not allowing equal time to alternative theories. What hypocrites!

Happy reading.

[RobotCzar]

Now that you have actually said something, let's examine what you say a bit more. Then we can decide who is dealing with dogmatic igorance. I have to say that you keep ignoring the points I make and simply repeat yourself or your "sources"

First off, you and your ilk love to say that people like me think all components sound the same. (I guess you think all components sound different, hence we end up with audiophile wood pucks, spikes, power cords, etc, etc.) "We" are quite clear that some components sound very different and make a "huge difference" in what one hears. And, guess what, people can tell them apart with their eyes closed.

I have repeatedly tried to tell you that there is no "time domain" and "frequency domain" that those concepts are made up to support the audiophile position. But, you are not listening, you simply assume you know the answer (gee, that sounds familiar).

One more time (no pun intended): Without time, there is no frequency--hence frequency is in the "time domain". I ask again, what is the "time domain"? You don't know, do you?

Phase errors ARE COMPLETELY connnected to frequency errors. If you have phase "errors" (i.e., distortion) you WILL have frequency errors and vice versa (the phase is beween frequencies). Now, we can measure high frequency performance of a amp (in terms of either frequency or phase) and see just how well an amp performs. Note that your article tips its ugly high-end hand in mentioning things like clarity and detail. Those are subjective, not scienitifc and technical terms. An electric signal carrying an analog of and audio signal, has time-varying characteristics such as frequency and amplitude (and differences in phase among frequencies) that can be measured. Those factors (along with distortion) relate to what we perceive as clarity etc.

Now, one last time: errors in phase MUST show up as errors in frequency response and noise (go ask someone who knows). The phase errors of modern home audio amps are so low that the distortion caused is below our ability to hear. One can measure the effect of "phase errors" by simply measuring the THD and frequency response using high frequency sine waves ("dynamic distortion"). Many cheaper amps DO have higher dynamic distortion than higher quality amps---we just can't hear it because it is still below the threshold of distortion we can hear.

"The quote refers to both the phase shift (time domain) and frequency response (frequency domain) performance of the power amplifier module, i.e audible tradeoff between the frequence domain performance ( 0.3dB at 20Hz) and the time domain performance (phase shift in the upper audio frequencies) and recommends a preferred tradeoff."

Phase shift and frequency are concepts invented by humans to understand time-varying signals (like sound). They are both aspects of the same phenomina and both are time based. They cannot be "traded off"--they are directly correlated. Higher phase distortion will result in higher frequency response errors (relative amplitude). You are being let down a path by people with an agenda. That path is that there are two unrelated entities "phase" and "frequency" and audio engineers only pay attention to one (how ridiculous). Get some real information, ask a real authority.

[JoeE SP9]

You're certainly not equating your experience with mine, now are you? And pray tell me, just what "experience" does an audio engineer (or a music lover either) have relating to the practical applications of power supply circuitry design for electronic components? Aren't you (conveniently) forgetting the fact that I spent the better part of 50 years actually working on the power supplies of both audio and video (television) products? It's probably not common knowledge, but the power supplies of just about all electronics products account for something like 75% of all servicing problems. So, I've spent a whole lot of time gaining actual "hands-on" experience in this area. Do you have any actual experience even remotely the equivalent? I'm not quoting from "theory" here although you seem to be implying that that is the case, while you on the other hand have experience to back up your assertions. Don't be offended, but I'm not inclined to buy it - especially because of some of the off-the-wall claims that I've seen you make here.



Au contraire, frere. It's not a bad analogy at all. I see that before you edited this post, you made a statement to the effect that " ... There is nothing stored in an amplifier (in)advance of playing music." Sorry guy ... that is totally incorrect (perhaps why you edited it out?). There is DC power stored in the filter capacitors of the unit's power supply at all times - before, during, and after music starts, stops, or whatever. The point of the analogy (which again, you conveniently ignore) is that the qualities of the gasoline hose cannot have any effect on the car's performance simply because it is doing nothing more than providing a raw material for the car to use at it sees fit. The AC power cord is fulfilling the exact same function ... it's providing a raw "material" for the electronic device to use as it sees fit - nothing more, nothing less.



Once again, your understanding of basic electronics is falling short of the mark. "Better" power cords contain filter networks all right, but I contend that they are there to hoodwink the gullible and impressionable among the unknowing, rather than to solve some "problem" of stray RF that might have a deleterious effect on the music. FYI, spark suppressors on car radios are not there to get rid of RF, but to suppress the electro-magnetic "noise" produced in the car's engine - and that is not "RF" by any stretch. I don't know whether it will relieve your mind (or just piss you off) to learn that the RF that you seem to think is such a problem for audio amplifiers that a special (and very likely expensive) after-market power cord is required to get rid of it, somehow magically disappears as the AC power is rectified and filtered by the power supply "inside the box".

P.S. I am not only a "music lover", but also a musician - with very sensitive ears. I have never in all of my life heard any RF emanate from a loudspeaker ever. Evidently, the power supply circuits in all of the audio amplifiers that I've ever heard must have done a quite remarkable job of eliminating it, don't you think?

I have in the past lived close to a ham radio operator who occasionaly would interfere with radio and tv reception for the whole block. I would also hear him through my rig. The same goes for a CB'er who was running an illegal linear amp.

[theaudiohobby]

RobotCzar,

You did not disappoint, ofcourse they are all inter-related, how can they not be, all these variables work together to produce a single entity called sound. Sound is a function of time but also of frequency. Frequency is also function of time, so it is any surprise that they are dependent variables. For the purposes of performance evaluation, there is a time domain representation and frequency domain representation, hence the terms. I found a very good reference for you that illustrates the issues, two statements in the link sum my points up very nicely.

Transforming a time-domain signal into its corresponding frequency-domain representation often helps to make apparent important characteristics of that signal.
...
Most musical signals have time-varying frequency content.

I hope this resolves the issues for you.

[musicoverall]

Those running a real statistical analysis would accept that you can hear a difference if you score better than would be expected by chance in one out of twenty cases of random guessing. ,

Upon rereading your post, please check my understanding of the above quote.

Are you saying that if I ran 20 sets of trials and scored better than chance (how many corrects out of 10 would better than chance be?) in just one set of trials, statistics would suggest that I do hear a difference? Even if I botched the other 19????

Sorry for being slow on the uptake here I do thank you for all your information.

[RobotCzar]

Upon rereading your post, please check my understanding of the above quote.

Are you saying that if I ran 20 sets of trials and scored better than chance (how many corrects out of 10 would better than chance be?) in just one set of trials, statistics would suggest that I do hear a difference? Even if I botched the other 19????

Sorry for being slow on the uptake here I do thank you for all your information.

No, I am sayting that if your score on one set of trials would only happen by chance 1 time in 20 (5% of the time), the accepted standard is to assume you were not guessing (i.e., your score was not due to chance). The point to note is that you could get a really good score and it could be lucky. So experiments need to be replicated.

If there are two alternatives (e.g. two different cables are being compared) then you would have a 0.5 chance of getting one trial correct. You would be expected to get 0.5 X 10 = 5 trials correct if you did 10 trials. So, how many would you get have to get right to convince a scientist? The number that would would get by chance only 5% of the time. Do you think you could calculate that? It is not trivial.

[theaudiohobby]

RobotCzar, you seem to have vanished into thin air.. Time to heckle you again since your sudden silence suggests that you now know that your original layman opinion of time domain and frequency domain representation was somewhat uninformed, embarrassed?

[RobotCzar]

If there are two alternatives (e.g. two different cables are being compared) then you would have a 0.5 chance of getting one trial correct. You would be expected to get 0.5 X 10 = 5 trials correct if you did 10 trials. So, how many would you get have to get right to convince a scientist? The number that would would get by chance only 5% of the time. Do you think you could calculate that? It is not trivial.

As it is not nice to propose a problem and not give an answer, here is mine:

Using the 11th row of Pascal's triangle (appropriate for outcomes with two alternatives such as an AB test) we see that there are 1024 possible combinations of 10 trials (2 to the 10th power). That row also indicates that the probibility of getting all right by chance is 1/1024 = 0.001 and the probablity of getting 9 right is 10/1024 = 0.01, 8 right is 45/1024 = .043.

Scientists generally accept your result as non-random if you do better than a probability of 0.05 by chance alone. Already, with 8 correct you are over the limit (.001+.01+.043 = .0531). Therefore, you must get 9 or more correct in ten trials to meet the standard criterion of significance. If you get 8, you are close and should probably run more trials or experiments. You should run more trials in any case as the more trials the more sensitive your test will be.

To sum up, if you do only 10 trials, a scientist would not be able to rule out that your score was due to chance (pure guessing) unless you got 9 or 10 correct. Formally, the null hypothesis (that you cannot tell differences) cannot be rejected unless your score is 9 or 10. If you do score a 9 or 10, you still could have been guessing, but the odds are so much against it that a scientist will believe that you were not guessing (this is a type 1 error).

[musicoverall]

As it is not nice to propose a problem and not give an answer, here is mine:

Using the 11th row of Pascal's triangle (appropriate for outcomes with two alternatives such as an AB test) we see that there are 1024 possible combinations of 10 trials (2 to the 10th power). That row also indicates that the probibility of getting all right by chance is 1/1024 = 0.001 and the probablity of getting 9 right is 10/1024 = 0.01, 8 right is 45/1024 = .043.

Scientists generally accept your result as non-random if you do better than a probability of 0.05 by chance alone. Already, with 8 correct you are over the limit (.001+.01+.043 = .0531). Therefore, you must get 9 or more correct in ten trials to meet the standard criterion of significance. If you get 8, you are close and should probably run more trials or experiments. You should run more trials in any case as the more trials the more sensitive your test will be.

To sum up, if you do only 10 trials, a scientist would not be able to rule out that your score was due to chance (pure guessing) unless you got 9 or 10 correct. Formally, the null hypothesis (that you cannot tell differences) cannot be rejected unless your score is 9 or 10. If you do score a 9 or 10, you still could have been guessing, but the odds are so much against it that a scientist will believe that you were not guessing (this is a type 1 error).

13 trials is my maximum, due more to happenstance than logistics or a desire to be statistically or scientifically correct. I have exactly 13 days in a row where my friend can swap (or not) the cables. Then I'd have to wait about 3 weeks to do it again. So I'm sticking with 13.

Thanks for the info.

[mystic]

13 trials is my maximum, due more to happenstance than logistics or a desire to be statistically or scientifically correct. I have exactly 13 days in a row where my friend can swap (or not) the cables. Then I'd have to wait about 3 weeks to do it again. So I'm sticking with 13.

Thanks for the info.

I see a problem with the experiment. Because your mood may change from one day to the next, your choices can be more a result of variations in how you feel than differences in the cables. This could result in the test having a negative bias. Given the method you will use, I don't know a way around the problem. It should not, however, in itself invalidate positive results.

If you decide to go ahead with the experiment, I recommend that you ask your friend to try to make the cable switches random -- his using coin flips should be good enough. You don't want have to deal with the possibility that he is trying to throw you a curve by using the same cable all 13 days, or giving you say 7 consecutive days of Cable A followed by 6 days of Cable B, when you assumed random switching at the start of testing.

Good luck!