[Roly1650]

ok... but your last statement is off track I think.  Sound changing to mono means something was broken, shorted, cross-wired etc.  That's very different than -75 db of noise.

Rubens experience is easy to explain when you know that all connectors in a usb cable are designed not to make or break together, coupled with the knowledge that the contacts are on one side of the plug only, so even with perfect alignment, there are contacts leading and contacts lagging. And with only slight misalignment in angular alignment left to right and/or vertically would cause loss of contact on some terminals and contact on others. The whole apparatus relies on the springs, stamped into the outer, to maintain contact and position, once the plug is all the way home. The usb plugs and sockets were designed this way to make hot swapping electrically safe. It's a mechanical system first and foremost, like any plugable/unplugable connector system.

A giant red herring.....

Did you actually read the Rane paper linked to? If you had, you'd realize that the document contained trivially easy solutions for rfi, (if you'd read the thread from the beginning, the first hypothesis) and ground loops, (introduced seamlessly and without explanation as the second hypothesis. Apparently though, it's everybody else that's unscientific). Sometimes it pays to read most of the thread rather than jump to conclusions based on only the last few posts.

 

[BiggerHead]

The "scientific" tests that castleofargh is requesting are fine, but nobody is realistically likely to do it, and that doesn't mean any further discussion or concern is unwarrented. Actually many things under the hood in science happen that way out of caution and lack of time to prove every non-global technical effect. The demands made here are actually higher than the demands made for considering an issue in any science lab. In fact, you can bet that in a lab, if there's a more expensive product that MIGHT POSSIBLY avoid some hypothetical risk, they will buy it if they have the budget. Of course they have career reputations on the line, not crisper cymbols.

I've heard plenty of mice, hard drives, video cards, etc, although mostly not through usb cables but jut on-board sound.

It is maybe slightly puzzling how a ground loop gets into the dac reference or amp section, but noise finds a ways. Even if there is a huge ground loop the current should mostly flow through the entire case of the DAC with the whole thing at one potential and all amplification relative to that and your headphones that you plug in also relative to that right? But noise is almost never that cooperative. You get some capacitive pickup somewhere etc. Yes it probably requires decent equipment to keep it out and it's much easier to stamp out good digitital parts with bad layout probably than the other way around.

I don't doubt that the paranoia is overplayed. I also don't doubt that for every person who had an obvious noise problem there is at least one who had a more subtle noise problem that yes, made their sound worse.

You had one card with obvious mouse noises. Do you just assume your next one where the mouse noise isn't obvious is fine?
It might be.

So you can buy an expensive DAC with known awesome noise isolation and/or replace your noisy computer,. On the other hand if there is a cheap solution that takes obvious to not obvious, and you know your noise isn't obvious, but who knows maybe could be better, then probably if you apply that cheap solution, now you know noise isn't a concern for you.

THIS actually is how scientists behave, not for things they are trying to prove, but for things they need to just work to be certain they won't get in the way. It's how most professionals behave actually.

All that said, I haven't seen a clear cheap solution proposed and while I might not demand the same level of testing that castle asked for, in order to care we would need at least one cheap solution with some testing. So far we see data for a $300 solution.

I actually think there is something of a consensus now. 1) Noise sometimes is a problem. 2) It "shouldn't" be. My part 3 is how do you know you're not in between? and my part 4 is if there is something cheap that reduces it 10 times and you don't know if you're in between, maybe it's not a bad idea to just use it. But we haven't proved there is that yet and I'm certainly not saying to fix a cheap dac with an expensive solution. It seems possible though that there is something to suit part 4.

 

[gregorio]

Honestly, some of you guys, this is the science forum for heaven's sake!
 
mmerrill99 was talking about ADCs not being sensitive enough and about RF and/or broadband ground noise leaking into cables and needing to be below -200dB to avoid human sensitivity to noise.
 
Let's put this into some kind of perspective: The most sensitive ADCs can resolve an analogue signal down to around -124dB, nowhere near the limit of 24bit (-144dB) because to achieve that limit would require the laws of physics to be broken. Down around -140dB, we are talking about the levels of noise created by sub-atomic particles colliding, the sound of electrons colliding inside say a single resistor. -200dB is another 1,000 times lower in level than this (!)
and roughly 1,000,000 times lower in level than your speakers or headphones can resolve.
 
Now if you want to run an ABX to prove to yourself that you can't hear any effects of noise which is 1,000 lower in level than sub-atomic particles colliding, then that's what you should do but obviously, we're very deep into audiophile "magic" territory here. I deal with the human perception of noise almost every day in my professional life, mostly in the -20dB to -70dB range. If those who engage my services thought I was wasting time on noise below about -80dB, they would fire me! Now maybe mmerrill99 is just having a good laugh at your expense or maybe he actually believes what he's posting but either way, please guys this is the science forum, not one of the "magic" forums, it's getting impossible to tell the difference. Just to be clear, I'm talking about analogue signals here, if we were talking about digital signals, the whole thing would be even more ludicrous than it already is!
 
G

 

[BiggerHead]

Gregorio, it's the science section but not everyone here is a scientist.  People make some statements that aren't quite on spot, ok.  Maybe even there can be some over-zealousness to find an explanation for something one believes they've heard (and maybe did). But there's a bunch that's been said that is reasonable too, and I've tried to put some perspective on that.  The OBVIOUS noise that two posters here(including myself and the mouse example above) have heard on computer related DACs is not -200 db.  The noise in the measurements shown is low, but also not -200 db. Again the idea that well there's junk DAC's but everything else is ok.. I don't think so. 
 
I've read some reputable posters here arguing from theory alone how every realtek dac is now an audiophile grade piece of equipment.  Well.. I'm here to tell you that I've got one on a good motherboard that absolutely results in less precise imaging as best I can describe it from one section of test music in particular, compared to an old mid-fi CD player, using the same bits, not that I'm shocked.  On certain pieces I could ABX it all day long (but I won't because what's the point in me proving to the world that my particular cheap computer DAC has sub-par sound, or it could be analog output cables! hah!).  If some computers have a lot of noise, why wouldn't others have a little?  What's the strong need to deny that possibility?  I guess 20db down from obvious is still marginally influential on perception, and that's 100 times less power. So there's a factor of 100 in quality in between obvious and still relevant.
 
Instead of picking a few overstated comments and dissmissing everything, why not just say.. ok, here's where some truth is, but it probably ends about there?  Then people can have a little debate and disagreement about that line is.

 

[mmerrill99]

ok... but your last statement is off track I think.  Sound changing to mono means something was broken, shorted, cross-wired etc.  That's very different than -75 db of noise.

Rubens experience is easy to explain when you know that all connectors in a usb cable are designed not to make or break together, coupled with the knowledge that the contacts are on one side of the plug only, so even with perfect alignment, there are contacts leading and contacts lagging. And with only slight misalignment in angular alignment left to right and/or vertically would cause loss of contact on some terminals and contact on others. The whole apparatus relies on the springs, stamped into the outer, to maintain contact and position, once the plug is all the way home. The usb plugs and sockets were designed this way to make hot swapping electrically safe. It's a mechanical system first and foremost, like any plugable/unplugable connector system.

A giant red herring.....

So explain how a digital signal that is interfered with in the way you describe results in a distant mono sound, not silence or choppy, buzzy sound. We will wait even though that herring is beginning to smell rather fishy 14 pages later

Did you actually read the Rane paper linked to? If you had, you'd realize that the document contained trivially easy solutions for rfi, (if you'd read the thread from the beginning, the first hypothesis) and ground loops, (introduced seamlessly and without explanation as the second hypothesis. Apparently though, it's everybody else that's unscientific). Sometimes it pays to read most of the thread rather than jump to conclusions based on only the last few posts.

Oh, the irony - I have participated in the thread from the beginning - you have joined in on page 14

I linked the Rane paper & never suggested that an expensive cable was needed to deal with RFI - I simply gave the technical explanations of why USB cables can make a difference. Unfortunately any technical knowledge is/was sadly lacking on this thread, preferring instead head-in-the-sand denials of anything technical.

 

[mmerrill99]

Reminds me of a recent thread on another forum.  It was asking if the hype about USB was overblown and the myriad problems of USB didn't indicate going that route was a mistake. 


The reasoning was you have these cables being so detrimental and goods one being necessary for 'good' USB.  You need these regulated 5 volt supplies.  You need this Regen, Intono or Corning optical links.  Connecting two devices via USB has become an entire cottage industry of its own.  You would get the idea USB for audio is fraught with horrid difficulty and mind really think it was just a bad idea.  Almost as if it were more fragile than LP rigs.  Which really are picky little beasts to keep in fine form, quiet and steady and optimally functioning.

On the other hand, no one can measure any benefits in the audio outputs with these.  Or in some cases these devices introduce some problems.  The designers of these things often have shaky theory or say they don't know it just works listen to it or admit they can't find a measured difference.  So we have neophytes fearing the USB.  Great gnashing of teeth on what best approach to take and hope USB is usable.  When in fact, asynchronous USB seems a wonderful way to largely isolate the computer from the audio system and get terrific measured and heard results. 

Somehow we have this burgeoning chicken little USB industry grown like mushrooms from a pile of .......shall we say anecdotal missteps. 

Well what is happening is that people are beginning to discover some issues with USB & computer audio in general which, if fixed result in far better sound. Most of these issues revolve around noise.
The situation is that computer audio can usually give very good sound & that is fine for most people - they are happy with it. But the audiophile hobby is always striving towards better sound & investigating areas of improvement & this is just the results of that striving & investigation

 

[mmerrill99]

Honestly, some of you guys, this is the science forum for heaven's sake!

mmerrill99 was talking about ADCs not being sensitive enough and about RF and/or broadband ground noise leaking into cables and needing to be below -200dB to avoid human sensitivity to noise.

Let's put this into some kind of perspective: The most sensitive ADCs can resolve an analogue signal down to around -124dB, nowhere near the limit of 24bit (-144dB) because to achieve that limit would require the laws of physics to be broken. Down around -140dB, we are talking about the levels of noise created by sub-atomic particles colliding, the sound of electrons colliding inside say a single resistor. -200dB is another 1,000 times lower in level than this (!)
and roughly 1,000,000 times lower in level than your speakers or headphones can resolve.

Now if you want to run an ABX to prove to yourself that you can't hear any effects of noise which is 1,000 lower in level than sub-atomic particles colliding, then that's what you should do but obviously, we're very deep into audiophile "magic" territory here. I deal with the human perception of noise almost every day in my professional life, mostly in the -20dB to -70dB range. If those who engage my services thought I was wasting time on noise below about -80dB, they would fire me! Now maybe mmerrill99 is just having a good laugh at your expense or maybe he actually believes what he's posting but either way, please guys this is the science forum, not one of the "magic" forums, it's getting impossible to tell the difference. Just to be clear, I'm talking about analogue signals here, if we were talking about digital signals, the whole thing would be even more ludicrous than it already is!

G

As usual you incorrectly attribute to me what I didn't say - I said "The difficulty in measuring it is not so much that it is below the noise floor of the ADCs used for measuring it - the difficulty is that ADCs don't have the dynamic range to accurately measure low level noise floor fluctuations in the presence of much larger multitone test signals."

You went of on another rant the moment I said that & as usual, for you, have your own version of what was said.

Look at what Biggerhead said just after that post - a notch filter is needed to take out the main signal in order that such low level FLUCTUATIONS might be measured. Why would that be correct, do you think? Why would we need to remove the main signal prior to trying to measure low level fluctuations with an ADC?

BTW, look at Rob Watts has to say about measuring noise floor modulation & ADCs - he unequivocally states that ADCs are not capable of measuring it. I would always take more note of waht is said by an actual audio designer involved directly in such measurements than an "armchair designer"

 

[ralphp@optonline]

Well what is happening is that people are beginning to discover some issues with USB & computer audio in general which, if fixed result in far better sound. Most of these issues revolve around noise.
The situation is that computer audio can usually give very good sound & that is fine for most people - they are happy with it. But the audiophile hobby is always striving towards better sound & investigating areas of improvement & this is just the results of that striving & investigation

Finally something we can agree on, namely that computer based audio can sound very good.
 
The high end audio world's fascination with and fixation on USB, a connection type that was designed to transmit data and not audio, when USB has been shown be to problematic has always puzzled me. There are other ways less problematic ways to get the digital audio data/signal from a computer to a DAC - coax and/or optical connection from the computer to the DAC, HDMI connection from the computer to the DAC, via Wi-Fi or Ethernet with the computer acting as a server. So many in fact that using a USB connections and then arguing about all the issues with the damn USB connections seems pointless.

 

[mmerrill99]

......

All that said, I haven't seen a clear cheap solution proposed and while I might not demand the same level of testing that castle asked for, in order to care we would need at least one cheap solution with some testing. So far we see data for a $300 solution.

I actually think there is something of a consensus now. 1) Noise sometimes is a problem. 2) It "shouldn't" be. My part 3 is how do you know you're not in between? and my part 4 is if there is something cheap that reduces it 10 times and you don't know if you're in between, maybe it's not a bad idea to just use it. But we haven't proved there is that yet and I'm certainly not saying to fix a cheap dac with an expensive solution. It seems possible though that there is something to suit part 4.

Well the way to test if you are in-between with noise (ie.e it's having an effect that you are not YET audibly aware of) is to borrow or buy an optical isolator & try it in your system. If it audibly improves the sound, you have a noise issue. Simple really.

 

[mmerrill99]

Well what is happening is that people are beginning to discover some issues with USB

Finally something we can agree on, namely that computer based audio can sound very good.

The high end audio world's fascination with and fixation on USB, a connection type that was designed to transmit data and not audio, when USB has been shown be to problematic has always puzzled me. There are other ways less problematic ways to get the digital audio data/signal from a computer to a DAC - coax and/or optical connection from the computer to the DAC, HDMI connection from the computer to the DAC, via Wi-Fi or Ethernet with the computer acting as a server. So many in fact that using a USB connections and then arguing about all the issues with the damn USB connections seems pointless.

Yes but that's not the subject of this thread & you could have saved both you & me so much keyboarding by saying this from the start instead of the arguments you put forth about digital

 

[BiggerHead]

 
Finally something we can agree on, namely that computer based audio can sound very good.
 
 

 
but this and the problems with usb cables for audio, not audio data, have been the discussion for several pages.   I agree, the thing was designed for data, but data only, not for use in noise sensitive analog equipment.   Using something else sounds smart me.

 

[BiggerHead]

By the way, how would coax or hdmi help with ground noise? Optical, of course, ethernet, probably. 

 

[mmerrill99]

 


Finally something we can agree on, namely that computer based audio can sound very good.

but this and the problems with usb cables for audio, not audio data, have been the discussion for several pages.   I agree, the thing was designed for data, but data only, not for use in noise sensitive analog equipment.   Using something else sounds smart me.

Well actually we need to take a step back & realise that general purpose computers are not designed within a noise budget that is in keeping with sensitive analog equipment & hence connecting such equipment to a computer results in these sorts of contortion to try to address the noise issue. It's far easier to deal with noise at source than to try to remove it downstream - at that stage it has often become embedded in the signal itself but even if it hasn't there is great difficulty separating noise from signals, in order to deal with it

 

[mmerrill99]

By the way, how would coax or hdmi help with ground noise? Optical, of course, ethernet, probably. 

Optical, yes - if good quality optical is used but even ethernet has issues i.e. the transformer doesn't block all noise.

 

[Ruben123]

I'm listening! It may be worth to add that it either does sound bad (mono or something like that: distant sound) or it sounds good.
I haven't heard anything inbetween, I'm using a $2 USB hub on a 1999 Compaq laptop.

What I don't get is: USB should have connection or not, me is told. So it should or should not output any sound. But actually there is an in between: namely the mono. But the question is how or why.