[dfkt]

You can get way better sound cards for cheaper. So it might not be deceptive per se, but why even bother with that NuForce sound card when you can get an ESI Dr. Dac, a Turtle Beach or even a humble Behringer, for less money - all of which perform in a professional manner. Just vote with your wallet...

 

[busyx2]

I am a car lover.  I used to look at the maximum HP and 0-60 numbers.  And I had very fast cars (0-60 < 5 sec).  Nevertheless, when I got to a point that I am looking for more space for baby seat and stroller, the 0-60 number became academic.  At the moment, No, I no longer care about stability of cornering at 100mph?  But do I care about day-to-day comfort and gas mileage?  Yes I do.
 
A triple 0 THD at 0DB is purely a measurement of academic performance.  Just like the top speed of a car that I owned will never be used or tested (if not I could be in Jail or face a $5K insurance premium).  Lexus woo their buyers simply with softer and more luxury seats and many customers do not drive the car in a manner of auto magazine editors.  
 
NWAVGUY, I applaud you, because your blog is very exciting.  Just like reading a car review, I got all the numbers, but if I am looking from a 'Consumer Report' angle, then I would rather want to know if these numbers are relevant to my experience.
 
http://en.wikipedia.org/wiki/Loudness_war   Quote:  "The standards of loudness would reach their limit in the 2000s. -10 dB had been the standard for the past several years, but this was often pushed to -9 dB. However, -6 to -5 dBFS is common in rock, contemporary R&B, pop, and hip hop music. Quieter exceptions to today's standards are rare. The releases of 2008 reached average levels as high as -3 dBFS, such as Angels & Airwaves' I-Empire, which yields almost 30 times the loudness of a THX standard recording (-20 dBFS).^{[citation needed]"}
 
So it's not true that recording are around 0dB, in fact the other way is true.
 
Also another paper:  http://www.kvglabs.com/PDF/DynamicRange.pdf
 
With MP3, the dynamic range could get as low as 20dB.
 
Quote:

 
I also do not understand how a 24bit DAC could have "too low" output level with "mp3s" (which are rendered to 16 bit) unless the scaling logic is grossly incorrect.

Finally, I can attest that many, if not most, modern pop music is mastered to 0 dBFS, in fact often more, having outright and quite noticeable clipping.


 

 

 

[vrln]

Thanks for the review, very good read. And I just bookmarked your blog as well There are nowhere near enough scientific reviews like yours in the audiophile world.

 

[Modo]

@nvawguy
How does the uDAC-2 compare to an integrated sound card, the kind it would replace in a semi-portable laptop rig?
 
 
Quote:

So it's not true that recording are around 0dB, in fact the other way is true.

Unfortunately, the wiki article talks about average levels per album, not about peaks. On "modern" recordings, you often get multiple peaks reaching exactly 0dB in every song. Normally loud music (rock, metal, etc.) can reach 0dB in most parts of an album (think every drum beat, sometimes even between beats). You can see it for yourself by using the waveform seekbar in Foobar. Some rock records look like people tried to make 0dB lines.

 

[nwavguy]

 
Quote:

Originally Posted by busyx2 /img/forum/go_quote.gif
 
A triple 0 THD at 0DB is purely a measurement of academic performance.  Just like the top speed of a car that I owned will never be used or tested (if not I could be in Jail or face a $5K insurance premium).  Lexus woo their buyers simply with softer and more luxury seats and many customers do not drive the car in a manner of auto magazine editors.  
 
NWAVGUY, I applaud you, because your blog is very exciting.  Just like reading a car review, I got all the numbers, but if I am looking from a 'Consumer Report' angle, then I would rather want to know if these numbers are relevant to my experience.
 
http://en.wikipedia.org/wiki/Loudness_war   Quote:  "The standards of loudness would reach their limit in the 2000s. -10 dB had been the standard for the past several years, but this was often pushed to -9 dB. However, -6 to -5 dBFS is common in rock, contemporary R&B, pop, and hip hop music. Quieter exceptions to today's standards are rare. The releases of 2008 reached average levels as high as -3 dBFS, such as Angels & Airwaves' I-Empire, which yields almost 30 times the loudness of a THX standard recording (-20 dBFS).^{[citation needed]"}
 
So it's not true that recording are around 0dB, in fact the other way is true.
 
Also another paper:  http://www.kvglabs.com/PDF/DynamicRange.pdf
 

 
I'm not trying to debate "triple 0" levels of THD. I agree that say .0007% THD is for all practical purposes the same as .007%. A case can even be made that .07% is still inaudible. But the uDAC-2 measures .7% at any setting of its volume control due to the internal DAC clipping problem. There are no zeros between the number 7 and the decimal point. As I've said, there's still considerable debate over what's audible as it depends on lots of variables, but many are documented as hearing 0.1% THD and, in certain cases, even 0.03% has been proven audible. But, overall, 0.5% is a more widely accepted standard of audibility. NuForce is over this level of audibility.
 
And your references above are the average level below 0 dBFS not the peak level. They're entirely different things. The higher the average level (which has been steadily pushed up over the years) the more the peaks hit 0 dBFS. Below is the music clip I used in the example for my review. It has a peak level of 0 dBFS (everywhere you see the blue lines hitting 100%) and an average (RMS) level of -12 dBFS. If you look at a track with an average level of -6 dBFS (and there are plenty around) it's "slammed" against 0 dBFS much of the time just as Wildstar reported. It's a fact the uDAC-2 will have ten to fifty times more distortion on each and every one of the peaks due to the way it's designed.
 
And you're confusing dynamic range as well. The lower the dynamic range (like the 20 dB number you quote) the more the signal will hit 0 dBFS as the recording is more compressed over a smaller total range.
 

 
Here's the popular Lady Gaga track "Just Dance" exactly as it's on the CD as shown in Audacity. Audacity is free. Anyone can download it and view their music as shown here. Below you can see what others have said about the music hitting 0 dBFS frequently. If you just put the CD in your drive and pushed play, this what is sent to the uDAC-2 and it will have what most consider high distortion for much of the track:
 

 

[nwavguy]

Quote:

@nvawguy
How does the uDAC-2 compare to an integrated sound card, the kind it would replace in a semi-portable laptop rig?

The internal sound hardware in laptops varies widely. So it's impossible to make a generalization. Some laptops certainly have better internal sound performance than the uDAC-2 can manage. Some, I'm sure are even worse. I've never seen even one, however, that clips 0 dBFS signals or has 10 dB of channel balance error like the uDAC-2 does. Generally, if you're happy with the sound of your laptop's headphone jack with your particular headphones, I wouldn't complicate things with an outboard DAC. And if you're not, I wouldn't buy a uDAC-2 but instead explore better options such as the suggestions I made in the review.
 

 

[bcwang]

Quote:

 
How about this: If NuForce wants to emphasize the sound of their products, I will attempt to make proper recordings of the sound of the NuForce uDAC-2, Behringer UCA202, and my Benchmark DAC1 Pre (as a reference) via the line outputs at a typical listening volume and I'll make excerpts available from my blog for downloading. That way anyone can listen for themselves and decide which DAC sounds best to them. I will carefully level match the signal from all 3 and record using the $1800 Benchmark ADC1 which is capable of extremely high quality transparent recordings. I'll document the exact set up so someone else can reproduce and verify what I did if they desire.
.......
 
So does this seem like an interesting idea? Are there Head-Fiers here who would like to download the files and listen for themselves?

I'd like to hear these samples.  Sounds like if you played back 44.1/16 on each one and used the benchmark ADC to record at 24/192 or at least 24/96, that would help.
 
-Same analog and usb cables
-Same output level for each source (if adjustable)
-provide the original reference sample as well if possible.

 

[holland]

Quote:

A triple 0 THD at 0DB is purely a measurement of academic performance.  Just like the top speed of a car that I owned will never be used or tested (if not I could be in Jail or face a $5K insurance premium). 
 
So it's not true that recording are around 0dB, in fact the other way is true.

That's incorrect.  0dBFS is not like the top speed of a car, so it's not a good analogy.  It is hit regularly unless you use "replay gain".  Is NuForce unique here?  I don't believe so.  I'd fathom clipping with 0dBFS is more common than not.  This is, in effect, why replay gain exists.  The loudness war, compressing the dynamic range, and then boosting the overall levels to clipping, is a major contributor.

 

[nwavguy]

Quote:

That's incorrect.  0dBFS is not like the top speed of a car, so it's not a good analogy.  It is hit regularly unless you use "replay gain".  Is NuForce unique here?  I don't believe so.  I'd fathom clipping with 0dBFS is more common than not.  This is, in effect, why replay gain exists.  The loudness war, compressing the dynamic range, and then boosting the overall levels to clipping, is a major contributor.

No it's not why replay gain exists, and 0 dBFS clipping in a DAC is very UNcommon. I challenge anyone to show me credible results of another device that clips in the same way the uDAC-2 does? I've probably measured close to 100 digital devices and I've never seen it before.
 
As I've tried to explain, 0 dBFS is the reference used in design, measurement, testing, etc. It's the cornerstone that everything else is built off of and referenced to. It's NOT normal for a DAC to screw it up. You can't reference to a "zero" level as it would be lost in the noise. And trying to use some intermediate value in the middle would create other inaccuracies and confusion. So 0 dBFS is the defacto standard and the uDAC-2 doesn't handle it well.
 
What I think you're getting confused is that some recordings clip at 0 dBFS. There are some ways certain software tries to "un clip" these recordings but the damage was already done when the CD was mastered. This has nothing to do with anything in the playback signal chain like the uDAC-2. It's done by the record label.
 
Replay gain exists because the average level varies a lot from track to track. Replay gain lets a software program analyze all the tracks and normalize (where possible) to a similar average listening volume as perceived by a human. There are lots of references on this. Please let me know if you need one?
 

 

[rymd]

I'm not a real big fan of your colour scheme and all the underlining+italicising+bold stuff, but this is all very very interesting indeed.

 

[holland]

Quote:

No it's not why replay gain exists, and 0 dBFS clipping in a DAC is very UNcommon. I challenge anyone to show me credible results of another device that clips in the same way the uDAC-2 does? I've probably measured close to 100 digital devices and I've never seen it before.
 
As I've tried to explain, 0 dBFS is the reference used in design, measurement, testing, etc. It's the cornerstone that everything else is built off of and referenced to. It's NOT normal for a DAC to screw it up. You can't reference to a "zero" level as it would be lost in the noise. And trying to use some intermediate value in the middle would create other inaccuracies and confusion. So 0 dBFS is the defacto standard and the uDAC-2 doesn't handle it well.
 
What I think you're getting confused is that some recordings clip at 0 dBFS. There are some ways certain software tries to "un clip" these recordings but the damage was already done when the CD was mastered. This has nothing to do with anything in the playback signal chain like the uDAC-2. It's done by the record label.
 
Replay gain exists because the average level varies a lot from track to track. Replay gain lets a software program analyze all the tracks and normalize (where possible) to a similar average listening volume as perceived by a human. There are lots of references on this. Please let me know if you need one?
 

I understand what you're saying.  At the same time, I can't fathom how NuForce would boost the signal by 1dB.  There's probably an anti-clipping mode in the chip, I don't know as I can't find the datasheets.
 
There should be nothing between the SPDIF receiver and the DAC.  What possible boosting can occur there?  I've looked at a few SPDIF receivers and I can't find a "boost" mode.  However, on DACs like the WM8741 there is an anti-clipping mode which drops the output level by 2dB.
 
I don't need a reference on replay gain.  Part of replay gain *is* for clipping and not volume adjustment track for track.  There is an album mode, which I would presume is more commonplace as that is what I use, to normalize an album not tracks.  The implication that replay gain exists for clipping is partially incorrect.  It is for clipping from compression artifacts.  As a side result 0dBFS will not be hit.  The de facto standard is 89dB, a good bit below 0dBFS for 16-bit.
 
Edit: can you open it up and let us know what chips are in use, or better yet a picture?  A google turned up the ES9023 for the DAC, but I can't find much of anything else on the interweb.
 
Edit:  Found the datasheet or not.

 

[nwavguy]

Quote:

I'd like to hear these samples.  Sounds like if you played back 44.1/16 on each one and used the benchmark ADC to record at 24/192 or at least 24/96, that would help.
 
-Same analog and usb cables
-Same output level for each source (if adjustable)
-provide the original reference sample as well if possible.

 
Thanks for the vote and your input. I agree with most of your comments but not the sampling rate. The fact is 95+% of the people listening won't be able to properly play back a 24/96 file let alone a 24/192 file. And those files would be many times larger than a 44/16 file which makes them much slower to download.
 
So rather than the vast majority of people having to sample rate convert the test files to 44/16, which may create its own artifacts and audible problems, I think it's better to record them in 44/16 so they can remain bit accurate just as the ADC1 captured them all the way to the DAC that's decoding them on the other end. I believe that will be more transparent.
 
And, if you're interested, extensive subjective testing has been done on Hydrogen Audio and elsewhere with respect to 24/96 vs 44/16 and the general consensus was there's no advantage for just recording and listening at 24/96. There is a advantage if you plan to edit/process the file as an intermediate step (such as a musician mastering their music) but not just for simple playback.
 
 

 

[dfkt]

Replaygain has nothing to do with clipping, it can't restore or recover anything that's been too loud to begin with. It just brings everything to a cozy perceived 89dB reference, be it track or album.

 

[holland]

Quote:

Originally Posted by nwavguy /img/forum/go_quote.gif
 
And, if you're interested, extensive subjective testing has been done on Hydrogen Audio and elsewhere with respect to 24/96 vs 44/16 and the general consensus was there's no advantage for just recording and listening at 24/96. There is a advantage if you plan to edit/process the file as an intermediate step (such as a musician mastering their music) but not just for simple playback.

Unless the DAC chip is configured a certain way, which requires upsampling to meet the configuration.
 
I would be interested in the samples as well.

 

[holland]

Quote:

Replaygain has nothing to do with clipping, it can't restore or recover anything that's been too loud to begin with. It just brings everything to a cozy perceived 89dB reference, be it track or album.

clipping prevention