DACS and "sampling rate sweet spots"

[RBFC]

We've heard many comments that some DACs have better performance at certain sampling rates. While it's been shown that often distortion products are higher in DACs running at 24/192 vs. 24/96, the situation surrounding sound quality of different sampling rates has not been so clearly demonstrated. What factors might contribute to better sound quality from 24/192 in some cases, and better sound in 24/96 in others?

Lee

 

[DonH50]

Interesting topic! My off-the-cuff quick answers (so probably wrong, but I'll start the ball rolling):

1. More bits imply greater resolution and dynamic range, meaning cleaner louder and quieter noise floor. I have personally found this more useful in the studio for recording and mixing than in the final cut, but have limited experience with hi-res output (e.g. BD).
2. Greater speed allows smoother filter roll-off to prevent in-band (audible) artifacts from steep filters. These artifacts include poor pulse response (ringing) and related phase shift in-band. Of course, higher rates come at a cost, usually in higher noise and distortion.
3. Trading distortion and noise for speed can thus be complex, involving not only the DAC core (actual converter) but amplifiers after and digital filters before. Optimizing competing parameters is non-trivial...
4. Because of those trades, it is typical to optimize the design for a given (target) sampling rate and bandwidth, and the compromises made among specifications may not be the ones you would have chosen.
5. Bear in mind that in many cases both the noise and distortion is well below the audible limit and/or masked by other factors, such as room acoustics and distortion in the speakers (which typically dwarfs all other distortion sources in a system).

I've no doubt left out a lot...

FWIWFM - Don

 

[Bruce B]

What factors might contribute to better sound quality from 24/192 in some cases, and better sound in 24/96 in others?

Lee

My experiences are that most DAC's have a sweet spot between a 44.1k and 48k integer. What I mean is that what I hear, some DAC's are better at 24/88.2 and 24/176.4 and are worse at 24/96 and 24/192

 

[garylkoh]

I agree with Bruce on this one. Some DACs are better implemented at 44.1k than 48k (may be due to clocks used? - one oscillator or two oscillators??)

However, I'm still curious because it might have been due to the music I have. I suspect that my DAC sounds better at 24/96 than it does at 24/88.1 but can't be absolutely sure because all the DAC can do is playback. May be the ADC's used was better at 24/96. When I re-sample 44.1 to 88.2 and 96 on my server, 88.2 sounds better than 96...... so for me the I am still not sure.

 

[DonH50]

Note resampling may have a major impact on the sound... An integer ratio (88.2/44.1) is much more straightforward and thus much less subject to various artifacts in resampling than a ratio of 96/44.1...

As with many things audio, and non-audio for that matter, it is sometimes very difficult to isolate just one parameter in the system... And even if you can, you may obviate the real-world interaction ("synergy") that takes place among components (for good or bad).

Edit: I keep getting tripped up by what audiophiles mean by "DAC" -- to me, a DAC is a single device/chip, one I have loved and hated over years of designing them, and not the whole box audiophiles call "the DAC". Please weigh my comments appropriately...

 

[Vincent Kars]

If I have to explain on computer audio forums what a DAC is, I say it is an outboard sound card

 

[microstrip]

I agree with Bruce on this one. Some DACs are better implemented at 44.1k than 48k (may be due to clocks used? - one oscillator or two oscillators??)

However, I'm still curious because it might have been due to the music I have. I suspect that my DAC sounds better at 24/96 than it does at 24/88.1 but can't be absolutely sure because all the DAC can do is playback. May be the ADC's used was better at 24/96. When I re-sample 44.1 to 88.2 and 96 on my server, 88.2 sounds better than 96...... so for me the I am still not sure.

Gary,
You are confusing me. Both you an Bruce state that some 44.1 x interger DACs are better implemented / sound better. And then you add " I suspect that my DAC sounds better at 24/96 than it does at 24/88.1but can't be absolutely sure because all the DAC can do is playback."

BTW, can you and Bruce put names on the DACs you are referring?

 

[Bruce B]

Gary,
You are confusing me. Both you an Bruce state that some 44.1 x interger DACs are better implemented / sound better. And then you add " I suspect that my DAC sounds better at 24/96 than it does at 24/88.1but can't be absolutely sure because all the DAC can do is playback."

What we're saying is some DACs sound better at 44.1k integers (88.2, 176.4, 352.8) and a few DACs sound better at the 48k integer (96, 192, 384). No DACs sound the best at both... even the Playback Designs, which I feel sounds better at 176.4/352.8

 

[garylkoh]

microstrip,

I'm still confused myself. I had always thought that my DAC - the Weiss Minerva - sounds better at 24/96 than 24/88.2. However, when I upsample redbook, 88.2 sounds better than 96. That is likely because of the integer vs non-integer. So, even in my own mind, it could just be that the music I have that is at 96 and 192 was better than the music I have that is 88.2 (and 176.4). I also had the Berkeley Alpha DAC, and on that, the 88.2 vs 96 was closer, but 96 still better.