Digital domain volume control

Status
Not open for further replies.
Jump to Latest
#42
You missed it ...

abraxalito said:
Where in what I've written have I made the assumption that its always 4 out of 16?
Click to expand...
:shhh:
Re-read your posts!

IAC ... and all the mostly pointless binary math aside ... George's orig. query is the topic at hand.

Search for "Spectral analysis and bit depth/resolution". Mags like Stereophile use the following procedure:

Quote:
I don't understand:
(a) how Stereophile derives bit depth from Spectral Analysis?

As a rule of thumb and all things being equal, each increase in digital word length lowers the noise floor by around 6dB. So if increasing the data word length fed to a DAC from 16 to 24 bits drops the noisefloor by 18dB rather than the expected 8x6=48dB, the DAC's own noisefloor is limiting resolution to 16 + 18/6 = 19 bits. (Which is still excellent, given that the current state of the DAC art is around 20 bits.)

Quote:
(b) why bit depth/resolution is randomly reported in Measurements (i.e., reported in review of a particular CD Player or D/A processor and not a particular CD Player or D/A processor in another review)?

I test all products with 16-bit data (the CD standard) and with 24-bit data (the maximum some products will handle). However, the limited space in any particular review means I can't publish all the graphs, just the ones that best illustrate the points I am making in the accompanying text.

John Atkinson
Editor, Stereophile
Click to expand...
 
#46
Can't find what was never hidden!

jan.didden said:
Found it!
This will be easy to follow, even by non-tech oriented types. Good explanations of pro's and con's of digital vol control.

From ESS no less:

http://www.esstech.com/files/3014/4095/4308/digital-vs-analog-volume-control.pdf
Jan
Click to expand...

But you missed this one altogether -- also from ESS/Martin Mallinson/RMAF 2011 presentation [I'm too lazy to read PDFs--YouTube to the rescue!]
https://www.youtube.com/watch?v=JYjHKv2_OqQ
Martin's second shorter presentation covers two topics that commonly arise in audio electronics. 1. Why does an analog volume control out-perform a digital one, and will a digital one ever measure up? We will look at noise sources and how an analog volume control needs to be designed to outperform even the best digital volume controls. 2. Why is jitter such a problem in digital audio devices? Can jitter actually be removed? We will go into some detail and describe the optimum clocking scheme for an Audio DAC, which is to have the DAC clock be the system master clock. But how can this be done if data is being delivered in another clock domain? We will explain what an Asynchronous Sample Rate Converter does and how it can help solve the clocking problem.
Click to expand...
 
#47
There was a link to that paper from ESS in post no6 in this very thread.. 😉

What they say is basically that done right it's hard to match a digital solution when it comes to the noise floor.

What they don't touch is that many DAC's have lower distortion at lower output levels, which is a win for digital. Reduce level in digital domain, analog output stage of DAC works in a more linear region = less distortion.

What they also show is that noise floor is constant in the DAC output signal when using digital volume control. They say this is reducing SNR and that's why audiophiles don't want it.

What they don't adress is that that noise is analog noise from the output stages and it's inaudible. You don't need to lower this noise floor when you lower the actual signal.
 
Last edited:
#48
hollowman said:
But you missed this one altogether -- also from ESS/Martin Mallinson/RMAF 2011 presentation [I'm too lazy to read PDFs--YouTube to the rescue!]
https://www.youtube.com/watch?v=JYjHKv2_OqQ
Click to expand...

Actually the paper I posted is the presentation you referred to - slides are identical.
If you check the YouTube you might actually see me sitting in that room 😉 so I didn't 'miss it' - I was there!

Jan
 
#50
georgehifi said:
Yes randy posted this up in the 6th post, and seems to be believable for 32bit dacs.

But nobody here or over at the other place backs up these claims by Malison from ESS


Cheers George
Click to expand...

Well it's not so much 'claims' as a reasoned explanation. It would be good if people finding an error in it would step up and say what the error is. It's not a matter of 'believing' it is right or wrong. If you can't find fault with it it would be logical to accept it.

jan
 
#51
It's correct to say that as a signal gets weaker by a digital level adjustment, the SNR is reduced.

What we need to understand though is that we do not need constant SNR for all actual levels of a signal.

How would that be for a signal at the threshold of hearing.. would that signal also need 120db SNR?

Of course not!

According to the flawed audiophile thinking a digital recording wold to be totally crap always for any signal that is not at the MSB. And guess what.. only DC could be... want to listen to DC? (or square waves as I mentined earlier.. square waves is alternating polarity DC) When a high level signal goes from positive to negative amplitude maxima, it will exercise all bits. For the majority of time the signal will have that **** poor SNR that you are afraid of.

We don't hear things the way "you" think.

The constant noise floor in the digital example of ESS is analog. The actual noise (quantization errors turned to noise by addition of dither before truncating to 24bit) that comes from doing math is way below that floor.
 
#53
A few years back when I was running an all active crossover system I had my doubts about 24 bit digital volume control done by the player software, so I ran extensive tests. THD, IMD, harmonic structure, noise. I found nothing but reduced S/N ratio, as was expected, Signal in = signal out with lower amplitude. Tested from 15Hz to 22kHz

In listening tests vs analog volume control I could here no difference, even tho analog measured worse.
Testing was done on headphones, passive crossover speakers and active. No audible difference.

A year after that I built an ESS DAC with volume control. Same results. I now no longer worry about it. If there are audible differences, something is wrong with the system.
 
#58
Pan said:
It's correct to say that as a signal gets weaker by a digital level adjustment, the SNR is reduced.

What we need to understand though is that we do not need constant SNR for all actual levels of a signal.

How would that be for a signal at the threshold of hearing.. would that signal also need 120db SNR?
Click to expand...

No argument - I'm not qualified. But doesn't this all assume that the gain structure of the system is reasonably optimized. The OP changed his speakers for a set that had much higher sensitivity. This situation resulted in a perception of "reduced dynamics" at "normal" volume. In this situation (or at least some possible situation) normal volume for listening becomes low volume for the digital volume setting. When the system gain structure was corrected for the higher sensitivity speakers. The perception of reduced dynamics was no longer.

Sheldon

edit: Maybe essential reading for setting up a system: http://www.diyaudio.com/forums/showthread.php?t=186018
 
Last edited:
#59
In this or any specific case, of course it can be a question of poor implementation which results in distortion and audible noise. A series of tests would be needed to find out about that.

What a high sensitivity speaker can do (or a high SPL capability rig, be it a beefy amp and/or sensitivte speakers) is play at such high levels so you actually hear 16bit dither noise.

Also amp noise floor will come out at a higher level with a more sensitive speaker.
 
Status
Not open for further replies.
Top Bottom