Digital Volume Control - understanding how it works.

[Nada]

Hi, only upstream of the pot.

Downstream of the pot, let's say you have a valve amp (pre, power, whatever) which has 60dB SNR at full power (maximum signal). Turn the pot down by, say 12 dB and you now have 60-12=48 dB of SNR.

Exacty the same thing happens with a digital pot, as far as everything downstream of the pot is concerned. The so-called resolution loss of digital is at the same rate, turn it down 12dB and resolution falls by 2 bits which is 12dB of SNR reduction.

The ideas above seem to me to ignore the central tenet of the first post as i understand it which is digital volume control is the only volume control that degrades the SNR at source. Using digital volume control makes the signal sink into the noise floor. Its a bit like having a record that is mastered at a very low volume so the music is quiet making the hiss, wow and crackle seem obtrusive ie a poor SNR.

Whether that's going to be audible to a listener in a particular system is another thing but lets at least understand the concepts so we can make an informed choice on what volume control we choose and not blindly accept ideas like "a free 8bits" or "48dB of pure attenuation with no sound quality loss" that are so common but do not seem to stand up to informed critique.

By the way if your preamp has

1. a SNR of 60dB

2.a volume control that only does attenuation of the signal while leaving its noise output at full level

you have an extremely poor preamp. Ive never come across one with such a flawed design.

Edited August 4, 2012 by Nada

[davewantsmoore]

I don't believe they (JRiver) recommend that at all. Where did you get that from?

Because it is the 'correct' way to do it..... Why would you reduce the digital signal before the DAC chip (using their software volume).... if you had the ability to do it inside the DAC or after the DAC ...... It doesn't make sense for the reasons I stated earlier.

Yes, their volume control is "perfect". 64bit digital attenuation is going to be mathematically transparent, this is not unique to JRiver..... but using digital attenuation in the wrong place (ie. when it could be used somewhere else which is smarter) .... is still not smart.

EDIT: Found something.

http://wiki.jriver.com/index.php/Volume

Internal Volume is useful in cases when hardware does not offer a volume, which is common with high-end sound cards and DACs.

Edited August 4, 2012 by davewantsmoore

[davewantsmoore]

[davewantsmoore]

Here is the entire presentation: http://www.esstech.com/PDF/digital-vs-analog-volume-control.pdf

[Dr X]

No Dave I don't think you found anything at all. Jriver do not distinctly recommend one method over another, rather that it's system dependent.

In my opinion, this is a complicated topic.

Sometimes turning up the hardware volume increases the noise you hear if you play silence (use Tools > Advanced Tools > Create Test Clips to generate silence files to test with).

If turning up the hardware volume causes you to hear more noise, it makes sense to not turn up that volume more than necessary.

But many times, turning up the hardware volume makes no difference to the noise in a stream of silence. So in these cases, maximize that volume and use Internal Volume so that Media Center has as much headroom as possible.

I test this by putting my ear right next to the speaker and adjust the volume while silence plays. With my Creative X-Fi hooked to a power amplifier, and also several more traditional Sony receivers I've tested, the hardware volume made no difference to the noise on the output. I don't know the deciding factor in this, but would like to.

So in those cases, Internal Volume should provide the best sound quality, especially if you do processing like Room Correction, Bass Management, Convolution, Linkwitz, etc. with Media Center where the extra headroom is useful.

[/Quote]

http://yabb.jriver.com/interact/index.php?topic=69988.0

Edited August 4, 2012 by Dr X

[Dr X]

Some more for you Dave:

http://yabb.jriver.c...p?topic=73602.0

Refer to Matt's (one of the developers of JRiver) comment below:

There's no single best answer...

[/Quote]

And some more:

This isn't a simple question. We were all standing around the white board arguing the merits of each approach, and basically decided "it depends." Since that's not too helpful, I'll at least post a few notes:

1) Media Center uses 64-bit precision for applying volume, meaning it's essentially perfect from the software side.

2) You lose 1-bit from your 24-bit output each -6dB. But since music has to be really loud to hear past 14 or 15 bits, this may not really matter. In other words, you lose more bits at low volume but you also need less bits at low volume.

3) The comparison between low-software-volume/high-hardware-volume and high-software-volume/low-hardware-volume depends on the hardware. Sometimes it's better to do hardware volume and sometimes it doesn't matter. One helpful test is to see if the background hiss (noise) at the speakers while audio is paused increases as you increase the hardware volume. If it does, this is an argument for not pushing the hardware volume more than you have to.

4) We couldn't think of any reason putting a resistor on the DAC would give a better volume than the internal volume. Resistors just throw bits away in the analog realm.

5) One tip is that I'd recommend the 'Volume Protection' feature in Media Center. It's designed exactly for cases where accidentally high volume could, as you said, "blow your face off."

[/Quote]

http://yabb.jriver.c...p?topic=67941.0

In summary I still have not read anything to back up your comment thet they JRiver recommend hardware analog volume over their Internal digital volume in all setups.

Edited August 4, 2012 by Dr X

[Nada]

Nice reference thanks Dave. Its useful for me to get some info to chew on.

For those who learn best from visual material Ive posted a nice representation of what digital volume control does from DAVE's ESS link with a 16bit file being digitally attenuated. The postulate of the OP proposes this happens no matter what the bit depth is 16/24/32...its the same SNR degradation but with a different depth of potential noise floor to work with:

Edited August 4, 2012 by Nada

[Dr X]

Here is the entire presentation: http://www.esstech.com/PDF/digital-vs-analog-volume-control.pdf

Correct me if I'm wrong, but their entire presentation is based around 16bit DAC's?

Practically all modern DAC's are 24bit which of course offers far more volume control without loosing bits.

[Dr X]

Remember the postulate of the OP states this happens no matter what the bit depth is 16/24/32...its the same SNR degradation:

So my analog pre-amp is disabled in my DEQX unit.

My digital attenuation in JRiver is at -96db.

Should I be hearing some noise? I can't!

[Nada]

Correct me if I'm wrong, but their entire presentation is based around 16bit DAC's?

Practically all modern DAC's are 24bit which of course offers far more volume control without loosing bits.

We dont listen to bits. Some of us think we can hear a poor SNR ratio and even find it detracts from our enjoyment of music. If you cant thats good for you. Enjoy.

Edited August 4, 2012 by Nada

[Nada]

So my analog pre-amp is disabled in my DEQX unit.

My digital attenuation in JRiver is at -96db.

Should I be hearing some noise? I can't!

I wonder if thats set too low for your system including hearing. Maybe try playing it quietly on a quiet night with digital versus analogue attenuation to see if the predicted digital artefacts and noise appear in your system?

Maybe try just listening to both in an uncritical way first. Do you find one more enjoyable?

Then listen to the background. Is one "blacker"? Then listen to something link a bell being hit. Does one sound distorted or tinny?

Theory for me is only there to help choose the right gear. Without theory we would need to listen to thousands of combos to work out wants best. Good theory can narrow down the experiment to what is feasible.

Please post your finding if you get to try it out.

Cheers.

Edited August 4, 2012 by Nada

[Dr X]

I wonder if thats set too low for your system including hearing.

But correct me if I'm wrong the article you linked to says that I will hear the full amount of noise my system produces, REGARDLESS of what attenuation setting I set. The reason I chose -96db is because it should deliver the lowest signal to noise ratio, yet I don't hear any noise.

FYI: I've compared analog to digital many times and I know which one I prefer subjectively.

Cheers.

Edited August 4, 2012 by Dr X

[davewantsmoore]

As you know, I agree that it depends....

I'm really trying to sit on the fence here, and explain the issue(s). I feel like I'm getting shoved off. Falling off a fence hurts. ;-)

Resistors just throw bits away in the analog realm.

I don't understand this comment.

Anyways. Their wiki isn't controversial: http://wiki.jriver.com/index.php/Volume

• In order to avoid the potential for clipping.... use their internal software volume if you use JR internal processing (REQ/convolution/etc)
  
• Unless you experience noise problems.... turn your system volume up .... and use some other volume control (either their software volume.... or inside your DAC ... or after your DAC)
  
• Internal (software) volume is useful if your external devices do not have their own volume control
  

Aside from audible noise issues they talk about .... the real crux of the matter for software volume (like JRs) will be: how much digital attenuation will you use before your DAC (using a software volume control) .... and how much will that affect the performance of your DAC ? (A small amount of digital attenuation before the DAC is not going to destroy performance).

[davewantsmoore]

Correct me if I'm wrong, but their entire presentation is based around 16bit DAC's?

Practically all modern DAC's are 24bit which of course offers far more volume control without loosing bits.

There are two separate issues.

Truncation/fractional errors, which add noise to the output. There are reduced with higher precision maths..... however keep in mind that the actual noise performance of (most) DACs sets a limit where over about 20bits of precision is swamped by the inherent noise performance of the DAC/circuit.

Then there is the issues raised in the OP. That is that when you digitally attenuate a signal you reduce the signal value, but the noise stays the same. You have reduced the SNR ..... When you adjust the volume in an analog way, you are reducing both the signal and the noise together. The slide Nada has quoted summarises this.

[Monkeyboi]

Thats way before my time Alan,

Maybe not? Just checked your profile and there is less than a decade differences in our ages. I would suspect (depending on your industry / professional background) you may very well have encountered electronic devices that used mercury wetted contacts but perhaps didn't or had no need to recognise it at the time.

so what youre saying is that electrons flow better in a liquid form than on solid state such as copper or silver??? :P LOL Wonder if they get dissy finding there way to gnd

No. The mercury used on mercury wetted contacts is there purely to significantly reduce the deterioration of the contacts conductivity over time by coating the contacts with Hg to prevent premature oxidation. The quality of the construction and materials used to make the contacts in the first place along with MTBF is what's important. Mercury wetting the contacts of some cheap and nasty switich or relay contacts won't magically make them better.

Cheers,

Alan R.

Edited August 4, 2012 by Monkeyboi

[davewantsmoore]

So my analog pre-amp is disabled in my DEQX unit.

My digital attenuation in JRiver is at -96db.

Should I be hearing some noise? I can't!

I'm not sure I understand your test.... What you need to do to test what I think you are testing is:

• Set JR to -96dB
  
• Amplify the analog signal coming out of your DAC to a normal listening level.
  
• Compare this output to digital volumes set at (much closer to) 0dB .... amplified to the same level.
  

If what you're saying is you turned JR down to -96 .... and that when set there the music would be output at a very very low volume) ..... and that you can't hear any background noise..... Then all that show is that it's hars to heard quiet sounds.

Sorry if I've missed something/misunderstood.

[Monkeyboi]

Don't think there is anyone doing mercury wetted switches now but I may be wrong.

Just one of many. http://www.pickeringrelay.com/dropdown/107series.html In this case a relay manufacturer.

Also - https://www.comus-intl.com/productpages/reed_switches_mercury_be.asp

https://www.comus-intl.com/productpages/mercury_wetted_relays.asp

Cheers,

Alan R.

[jerry20]

I have no idea.

[sir sanders zingmore]

I'm not sure I understand your test.... What you need to do to test what I think you are testing is:

• Set JR to -96dB
  
• Amplify the analog signal coming out of your DAC to a normal listening level.
  
• Compare this output to digital volumes set at (much closer to) 0dB .... amplified to the same level.
  

If what you're saying is you turned JR down to -96 .... and that when set there the music would be output at a very very low volume) ..... and that you can't hear any background noise..... Then all that show is that it's hars to heard quiet sounds.

Sorry if I've missed something/misunderstood.

Maybe the point is that with JR at -96 the SNR is at its absolute worst. If you can't hear any noise when it's at it's worst then there are no worries about using digital volume control

[Nada]

Maybe the point is that with JR at -96 the SNR is at its absolute worst. If you can't hear any noise when it's at it's worst then there are no worries about using digital volume control

-96 is so low I doubt there will be anything much to hear at all

a listening test needs some music to listen to

just testing for noise without music is a furphy in my current view

[Nada]

But correct me if I'm wrong the article you linked to says that I will hear the full amount of noise my system produces, REGARDLESS of what attenuation setting I set. The reason I chose -96db is because it should deliver the lowest signal to noise ratio, yet I don't hear any noise.

FYI: I've compared analog to digital many times and I know which one I prefer subjectively.

Cheers.

if you are going to test music you need to hear something I suspect?

if you turn the music down so low you hear nothing your SNR=0 subjectively which is not useful

for you to understand these issues I suggest you read my advice to you above and follow it exactly - listen when its very quiet so the ambient room noise is <20dB, listen at low volume etc otherwise you are just spinning words

I predict you will still not hear a difference with your system that includes your brain, other then to reinforce your current position, but we never know.....

Edited August 4, 2012 by Nada

[davewantsmoore]

Maybe the point is that with JR at -96 the SNR is at its absolute worst. If you can't hear any noise when it's at it's worst then there are no worries about using digital volume control

Yes, thats' the point. But when you check to see if you can hear the noise, you have to have the output from the DAC amplified to normal listening levels... otherwise the test is meaningless.

I predict you will still not hear a difference

I wouldn't be surprised either.

Edited August 4, 2012 by davewantsmoore

[Dr X]

Maybe the point is that with JR at -96 the SNR is at its absolute worst. If you can't hear any noise when it's at it's worst then there are no worries about using digital volume control

Exactly, but of course if the amount of noise is constant as suggested by the OP...that would mean that I'd hear exactly the "same level" of noise at any and all attenuation levels.

-96db was just an example, I should hear exactly the "same level" of noise at -48db and 0db.

Edited August 4, 2012 by Dr X

[Dr X]

I predict you will still not hear a difference with your system that includes your brain, other then to reinforce your current position, but we never know.....

I think you created this entire thread to re-enforce your current position. Have you performed an experiment to hear noise brought about by digital attenuation? Please let me know in case I missed it.

Edited August 4, 2012 by Dr X

[Dr X]

-96 is so low I doubt there will be anything much to hear at all

a listening test needs some music to listen to

just testing for noise without music is a furphy in my current view

That's just the thing, I can't hear any noise with or without music playing at ANY attenuation level!