[chongky]

I've just bought for myself the O2/ODAC combo and is loving every minute of it. I'm hearing more low-level details at quieter passages of music, and I found myself lowering the volume knob to enjoy the tonal colors of well recorded CDs. However, according to NWAvguy, there is no benefit to a SNR higher than 110 dB. I'm no engineer, but I thought I read somewhere it is possible to listen to well below the noise floor? Now that many Dacs like Xonar Essence One, Benchmark Dac2 etc. (most utilizing the highest quality DAC chip from Sabre or TI) offer SNR exceeding 120 dB, I'd like to ask if this does make a difference? Or is it just our (my?) delusion that better specifications lead to better sound?

 

[Strangelove424]

Whatever the dynamic range of the equipment is, your hearing system still has dynamic range limits of its own. You can look at this chart for a reference.
 

Start with the assumption that, unless you live in an anechoic chamber, your ambient is probably no less than 30db (that's considered ambient noise for a quiet house in the country). Above 120db you reach pain levels. At that point your ears will adapt to loud noises, tighten up the anvils and stirrups, and thus minimize the increased loudness. Your ears have their own loudness normalization. So everything above that 120db is pretty much useless to you, you're just begging for pain and permanent hearing damage. Subtract 120db from your 30db ambient floor and realistically you are looking at a 90db SNR max. And even that would be hard on your ears occasionally. 70-80db is probably a more realistic number for an enjoyable SNR, more akin to attending a classical concert. Not saying it's bad for equipment to have high SNR, just saying your ears only have so much dynamic range themselves, and at a certain point it does indeed become pointless, perhaps even painful. 

 

[chongky]

Nice graph, and very lucidly argued. Makes me wonder why people are so intent on purchasing HD tracks when well-recorded (and well mastered) CDs deliver 120dB with noise-shaped dither.
 
I was just now reading a review of the NAD M2 by The Absolute Sound (not the most scientific of mags) and came across this statement:
 
"if we start with a CD player with a SNR of 115dB, feed its output to a preamplifier with a SNR of 108dB, and then drive a power amplifier whose intrinsic SNR is 115dB (all great specs), the system’s overall SNR is only 84.1dB referenced to 1W (all SNR numbers are un-weighted)."
 
How Robert Harley came to the figure 84.1dB I don't know, but with a SNR > 120 dB at the DAC side it does seem to allow audio degradation along the pre-amp and power amp?
 
Benchmark is now selling a power amp with the purported SNR of > 130dB so it does seem that audio degradation do occur as the signal gets amplified...

 

[Strangelove424]

Amping the signal will lower the SNR, it always adds some noise to the signal, just because of physics, no amp can avoid that. My soundcard has dac SNR of 124db, amp SNR at 110db. At normally listening volumes it is black.  However, if I turn the AC off, fans off, lower fan rpm on my computer, close windows, and and just sit there in a zen silence, I can just barely hear a faint noise floor with the volume maxed. The rush of wind is louder than that ridiculously faint noise floor. What happens if I push the play button at that volume? The pain would be horrendous. 
 
What if theoretically there was an amp with 200db SNR? It could play a 200db signal perfectly clean! It would be the best amplifier in the history of amplifiers... except for the fact that anyone who experienced that SNR in person would lose their hearing. They would be wiping eardums off the amp between shows. At a certain point (and it's pretty easy to get there nowadays) the SNR is beyond the ability of human beings to safely enjoy. 

 

[stv014]

Originally Posted by chongky /img/forum/go_quote.gif
 
"if we start with a CD player with a SNR of 115dB, feed its output to a preamplifier with a SNR of 108dB, and then drive a power amplifier whose intrinsic SNR is 115dB (all great specs), the system’s overall SNR is only 84.1dB referenced to 1W (all SNR numbers are un-weighted)."
 
How Robert Harley came to the figure 84.1dB I don't know, but with a SNR > 120 dB at the DAC side it does seem to allow audio degradation along the pre-amp and power amp?

 
While the specs provided are not enough to explain to exact figure of 84.1 dB, the main reason why it can decrease so much is that the power amplifier is not run at its maximum output level (i.e. the 115 dB is very likely referenced to the highest voltage it is capable of without clipping, while the 84.1 dB is at 1 W power), and that there is significant gain after the volume control. If the volume control is in the pre-amplifier, and the system has excess gain (it would clip at full volume), then the SNR can indeed be relatively low at a low output level.
 
However, this would not be fixed, or even significantly improved by increasing the SNR of the CD player. Generally, in any decent playback system with a digital source, noise that is added before the volume control is rarely the problem. With music that does not have very high dynamic range, it could even be worse than "CD quality" 96 dB, and still not be audible.
 
By the way, 84.1 dB unweighted SNR (which would be ~87 dB with A-weighting and white noise) is not necessarily audible at 1 W power with speakers. Assuming a sensitivity of about 90 dB/W, it translates to a few dB of A-weighted noise SPL, and that can be masked by typical ambient noise levels.

 

[stv014]

  My soundcard has dac SNR of 124db

 
Actually, when the sample rate is a multiple of 44.1 kHz, the Xonar Essence STX has much lower SNR, "only" about 110-111 dB, which is of course still good enough for a line output. The advertised figure is at 48 (or 96) kHz sample rate, and even then it might be 118-120 dB in practice, perhaps depending on other parts of the PC. Lowering the volume on the card obviously reduces it further, since the volume control is digital.

 

[Strangelove424]

  Quote:

   
While the specs provided are not enough to explain to exact figure of 84.1 dB, the main reason why it can decrease so much is that the power amplifier is not run at its maximum output level (i.e. the 115 dB is very likely referenced to the highest voltage it is capable of without clipping, while the 84.1 dB is at 1 W power), and that there is significant gain after the volume control. If the volume control is in the pre-amplifier, and the system has excess gain (it would clip at full volume), then the SNR can indeed be relatively low at a low output level.
 
However, this would not be fixed, or even significantly improved by increasing the SNR of the CD player. Generally, in any decent playback system with a digital source, noise that is added before the volume control is rarely the problem. With music that does not have very high dynamic range, it could even be worse than "CD quality" 96 dB, and still not be audible.
 
By the way, 84.1 dB unweighted SNR (which would be ~87 dB with A-weighting and white noise) is not necessarily audible at 1 W power with speakers. Assuming a sensitivity of about 90 dB/W, it translates to a few dB of A-weighted noise SPL, and that can be masked by typical ambient noise levels.

 
If that is the case with the quoted numbers on the amp in Absolute Sound, and the specs are based on voltage swings, then the SNR or max power output spec is flawed to begin with there. That's not the first time I saw a magazines pick up a company's flawed or misleading specs and quote them like measurements.

 

[Strangelove424]

   
Actually, when the sample rate is a multiple of 44.1 kHz, the Xonar Essence STX has much lower SNR, "only" about 110-111 dB, which is of course still good enough for a line output. The advertised figure is at 48 (or 96) kHz sample rate, and even then it might be 118-120 dB in practice, perhaps depending on other parts of the PC. Lowering the volume on the card obviously reduces it further, since the volume control is digital.

 
I only keep line out at 50% so the amp can stay at ~50%. In my experience, these high signal levels are ridiculously loud, and the audibility of the noise floor at anything close to listening levels isn't something to be concerned with.     

 

[bigshot]

  Makes me wonder why people are so intent on purchasing HD tracks when well-recorded (and well mastered) CDs deliver 120dB with noise-shaped dither.

 
Welcome to the club!

 

[blades]

  I've just bought for myself the O2/ODAC combo and is loving every minute of it. I'm hearing more low-level details at quieter passages of music, and I found myself lowering the volume knob to enjoy the tonal colors of well recorded CDs. However, according to NWAvguy, there is no benefit to a SNR higher than 110 dB. I'm no engineer, but I thought I read somewhere it is possible to listen to well below the noise floor? Now that many Dacs like Xonar Essence One, Benchmark Dac2 etc. (most utilizing the highest quality DAC chip from Sabre or TI) offer SNR exceeding 120 dB, I'd like to ask if this does make a difference? Or is it just our (my?) delusion that better specifications lead to better sound?

 
It isn't a delusion.  It is just a matter of misunderstanding.  Better specifications can certainly indicate better performance.  But when the specifications are below the level of audibility, they lose that capability.  Every thing in audio is measurable.  Not everything is audible.
 
Example.  One amplifier has a THD spec of 1/10 percent and another has a spec of 1/100 percent.  Does the latter one perform better?  Yes.  Does it sound different?  No, because both specs are below the level of audibility.

 

[AnAnalogSpirit]

  Amping the signal will lower the SNR, it always adds some noise to the signal, just because of physics, no amp can avoid that. My soundcard has dac SNR of 124db, amp SNR at 110db. At normally listening volumes it is black.  However, if I turn the AC off, fans off, lower fan rpm on my computer, close windows, and and just sit there in a zen silence, I can just barely hear a faint noise floor with the volume maxed. The rush of wind is louder than that ridiculously faint noise floor. What happens if I push the play button at that volume? The pain would be horrendous. 
 
What if theoretically there was an amp with 200db SNR? It could play a 200db signal perfectly clean! It would be the best amplifier in the history of amplifiers... except for the fact that anyone who experienced that SNR in person would lose their hearing. They would be wiping eardums off the amp between shows. At a certain point (and it's pretty easy to get there nowadays) the SNR is beyond the ability of human beings to safely enjoy. 

I apologize beforehand on two counts: one, not reading through the thread completely due to my excitement at finding intelligent conversation, and two, liking BOTH high SNR aka "cleanliness" and "noisy" vintage tube sound as well as dead silent noise floor from solid state gear, whether it be any variation of tonality or just plain reference, of which there are probably as many variations as there are users on Head-Fi and BEYOND... too many to classify? Dunno, but it sure is fun to listen, and try? xD. Nice name Strangelove424, I'm really enjoying my NOS HD414SL through a SENSE G3 with NOS tubes, The description is detailed in my User Profile (I don't want to threadcrap too much), but I really am enjoying how CLEAN and REFERENCE they sound when its REALLY QUIET around my rather noisy surrounding abode. I often listen to electronica and acoustic/ambient at 120dB+ when i can get away with it... at 16bit Lossless FLAC. Sorry this isn't an equiment or format thread this is SOUND SCIENCE, I should really mind my manners, I'm just excited to see a good conversation, I haven't been on Head-Fi in a long time and I'm hoping this thread moves in a positive direction...

 

[Strangelove424]

  I apologize beforehand on two counts: one, not reading through the thread completely due to my excitement at finding intelligent conversation, and two, liking BOTH high SNR aka "cleanliness" and "noisy" vintage tube sound as well as dead silent noise floor from solid state gear, whether it be any variation of tonality or just plain reference, of which there are probably as many variations as there are users on Head-Fi and BEYOND... too many to classify? Dunno, but it sure is fun to listen, and try? xD. Nice name Strangelove424, I'm really enjoying my NOS HD414SL through a SENSE G3 with NOS tubes, The description is detailed in my User Profile (I don't want to threadcrap too much), but I really am enjoying how CLEAN and REFERENCE they sound when its REALLY QUIET around my rather noisy surrounding abode. I often listen to electronica and acoustic/ambient at 120dB+ when i can get away with it... at 16bit Lossless FLAC. Sorry this isn't an equiment or format thread this is SOUND SCIENCE, I should really mind my manners, I'm just excited to see a good conversation, I haven't been on Head-Fi in a long time and I'm hoping this thread moves in a positive direction...

 
Glad you found the discussion informative. Regarding "clean" vs" noisy" I have found that what people refer to as "pleasing" (and what I sometimes prefer listening to myself) doesn't always have to have the cleanest SNR ratio. CD's dynamic range is about 90db and Vinyl about 70db, with a far higher noise floor. Neither can take full advantage of 120db SNR, but even so, why is it that people often prefer vinyl? By all technical considerations, vinyl is a lower fidelity format than CD with a much higher noise floor. The noise floor is so high that one can hear the noise floor coming in first before the music. Once the music arrives, the noise floor vanishes beneath it, but I cannot help but wonder if it's that higher noise floor that explains why people (myself included) enjoy listening to vinyl. The higher noise floor could very well explain why vinyl sounds "warmer, fuller, more rounded", etc. Same can apply to some tube amps with high levels of noise. I find it interesting that we struggle for clean signals, high DRs and SNRs, and then decide to run noisy material through that crystal-clean system. A part of me wonders if human beings need some level of background noise for something to sound natural or enjoyable.  
 
BTW, listening to music at an SPL of 120db (even acoustic or ambient) will really take its toll on your hearing. Be careful. 

 

[Dark_wizzie]

   
Glad you found the discussion informative. Regarding "clean" vs" noisy" I have found that what people refer to as "pleasing" (and what I sometimes prefer listening to myself) doesn't always have to have the cleanest SNR ratio. CD's dynamic range is about 90db and Vinyl about 70db, with a far higher noise floor. Neither can take full advantage of 120db SNR, but even so, why is it that people often prefer vinyl? By all technical considerations, vinyl is a lower fidelity format than CD with a much higher noise floor. The noise floor is so high that one can hear the noise floor coming in first before the music. Once the music arrives, the noise floor vanishes beneath it, but I cannot help but wonder if it's that higher noise floor that explains why people (myself included) enjoy listening to vinyl. The higher noise floor could very well explain why vinyl sounds "warmer, fuller, more rounded", etc. Same can apply to some tube amps with high levels of noise. I find it interesting that we struggle for clean signals, high DRs and SNRs, and then decide to run noisy material through that crystal-clean system. A part of me wonders if human beings need some level of background noise for something to sound natural or enjoyable.  
 
BTW, listening to music at an SPL of 120db (even acoustic or ambient) will really take its toll on your hearing. Be careful. 

Can't you devise a little test to test this idea? Play a normal CD track and add noise to it?

 

[Strangelove424]

I've always wanted to run a test like this. I would start by telling all the participants they are there for a blind test of different amps, some solid state and some tube. Their objective would be to rank the amps in order of preference and then state whether they thought each amp was SS or tube. In fact, the participants would be listening to the same neutral SS amp over and over with different db levels of noise added in from a separate track. Perhaps you could get even more complex with it, and use noise from different bands in addition to differing levels. Since the noise would be piped in from a separate track, all parameters of it could be controlled, including start time. So the administer of the test could fade in the noise to the correct level after the music starts, so there is no obvious background noise when the music is not present. It has always been a wish of mine to conduct such a test or atleast read about the results of one.

 

[castleofargh]

I don't know if independent noise can do the trick. I would guess it still has to be somehow correlated to music, or focused at least. maybe only on bass to make for some "texture" ^_^ .
but the idea that the cleanest sound isn't the best sounding music, I very much believe that. at least as long as we listen to crap masters I'm sure of it. some stuff sound great in a car, but when I get them in my headphone I cry. I've been disappointed a good number of times like that.