Digital Volume Control - understanding how it works.

[Nada]

Digital Volume Control and SNR by Sean Adams http://forums.slimdevices.com/showthread.php?t=30916&page=2

" I think the easiest way to understand this is to forget about numbers, decibels, and bits per sample for a minute, and just think about what's coming out of the DAC.

To oversimply only slightly: there are two things always coming from the DAC. 1) signal and 2) noise.

The level of the noise output stays the same no matter what signal level is being produced. That is really important to understand!

When the DAC is making a loud signal, there is a lot of sig

nal and a little noise. That's a high SNR, which is good.

However, when the DAC is making a quiet signal, you have a little signal and a little noise. If we now consider the noise level in relation to the signal level, the noise is now louder. The noise level hasn't gone up in absolute terms (eg volts), but relative to the signal it has, so you now have a bad SNR.

Now consider a simple resistor attenuator being fed by a loud (good SNR) signal from the DAC. When the voltage passes through the resistor divider, everything gets attenuated - the signal and noise together. You have the same* SNR coming out of the divider as you had going in, i.e., the DAC's optimal SNR is preserved.

OK, now back to bits per sample. As you can see, the above effects really don't have much at all to do with bits per sample. We could send a million bits per sample, and it would still be the same. So why does bit depth matter? What is the significance of 16 vs 24 bit?

What matters is that we send enough bits per sample that the DAC's full dynamic range is utilized. It is important to realize that the DAC's dynamic range is finite, and is less than its input word size - more like 20 bits, since it is limited by its output noise level.

By "expanding" a 16 bit signal to 24 bit, all we are doing is saying "these 16 bits go in the most significant slots of the 24 bit word". We haven't improved the SNR of the signal, any more than you can "enhance" a digital photo the way they do on CSI.

If we attenuate the 16 bit signal, yes, the zeroes and ones will migrate down into the least significant bits of the 24 bit word, and yes, if we still "have all the bits" we could then mathematically go in reverse and get back to the same data. But that is not what the DAC does with the signal! The bits represent a smaller signal now than they did before. We still have exactly the same decreasing SNR effect.

Sending 24 bits into the DAC just means we aren't making it any worse than it already is. We haven't "bought more headroom"... it does NOT mean that those first 8 bits of attenuation are "free".

To prove this, you could play a sine wave through the DAC and measure the SNR at each volume step. We would expect to see the SNR decrease as the volume is decreased. If there were anything special about the point where we start "losing bits", or if we were really getting "extra headroom", then the plot would decrease slowly (or not at all) until it reaches that point, and then there would be an inflection.

However, that is not what you'll see. The SNR will simply decrease with the signal level, all the way down.

I hope this helps... for extra credit maybe someone will try testing this?

Actally, there are a number of secondary effects which reduce the SNR by the time it gets through the amplifier, but these are vanishingly small in comparison.

[Arg]

Your point is....?

[Lil Caesar]

...happy with the way my NAD M51 does digital attenuation. Whether the 35 bits helps in terms of noise vs SNR or not is hard to tell... After all we are talking about stuff happening at 0.01 and less of 1 percent....

Edited July 27, 2012 by Lil Caesar

[kdoot]

The other issue is that in manipulating the digital samples to reduce their amplitude, you create additional noise and artefacts as part of what Sean Adams is calling "signal". It's not noise created by the DAC, but it's still noise.

With a 24 bit DAC, your digital volume control can reduce the amplitude of this digital error noise by up to 256 times compared with 16 bit!

Dithering, by the way, seeks to randomise the digital attenuation errors so that they end up more like white noise than harmonics of the signal. You really need dithering to listen on a 16 bit DAC with digital attenuation, but I find on a 24 bit system I prefer it undithered.

[Nada]

Your point is....?

This explanation debunks the endemic computer audiodefiled myth that that digital volume control can provide transparent attenuation.

The myth goes like this:

Take a 16bit file and change the bit depth to 24bits on the player output and then run it out via a 24bit capable signal path to a 24bit DAC (or a 24bit file along a 32bit path.) That creates an extra 8 bits for depth. Then we can attenuate the volume by 6dB x 8bits = 48bits with NO LOSS. This means there is a totally transparent no loss no mess no artefact volume control that is the only free and noise free volume control in the universe. Wow.

If you haven't been fooled by this myth you've done very well. Like I said its endemic.

What this article explains better then Ive ever been able to is that digital volume control results in the signal sinking into the noise floor with a loss in dynamic range.

That's extremely important for any system planning with respect to choosing volume control.

Preamps still have a valid place. Why? To put it simply they attenuate both the noise and the signal preserving the SNR and range.

...happy with the way my NAD M51 does digital attenuation. Whether the 35 bits helps in terms of noise vs SNR or not is hard to tell... After all we are talking about stuff happening at 0.01 and less of 1 percent....

The explanation above points out it essential doesnt matter whether you use 16bits or 24bit or 35bits. Youve got the same problem.

Dont believe any theory I say.

I think its sensible to test it out ourselves.

To test connect directly without a preamp and use a software player to put out 16bit/24bit/32bit (hopefully your transport-DAC will accept those bit depths).

Listen when its really quiet and turn the digital volume down a fair way e.g. -50dB and listen to the different bit depths.

Do you hear any difference?

Then slot in a good preamp or attenuators and use no digital attenuation.

What sounds better?

The other issue is that in manipulating the digital samples to reduce their amplitude, you create additional noise and artefacts as part of what Sean Adams is calling "signal". It's not noise created by the DAC, but it's still noise.

With a 24 bit DAC, your digital volume control can reduce the amplitude of this digital error noise by up to 256 times compared with 16 bit!

Dithering, by the way, seeks to randomise the digital attenuation errors so that they end up more like white noise than harmonics of the signal. You really need dithering to listen on a 16 bit DAC with digital attenuation, but I find on a 24 bit system I prefer it undithered.

Thats excellent info thanks. I wonder how loud the undithered harmonics can be ?

I also wonder what's the difference in volume error using 16bit versus 24bit is? Can you check my working out as I will probably get this wrong:

I guess using 16bits there is loss of accuracy in the last bit (is that called the LSB?).

I think that reduces the integrity to 15 bit?

That will give 2 to the power of 15 discreet volume steps = 32768 over a dynamic range of 6dB x 15 = 90dB or 0.00274658203125dB per step, if my mental arithmetic is OK? I dont know if my hearing is up to tell the difference between 0.003dB steps. Does this mean 16bit encoding is enough for dB steps?

[bhobba]

Do you hear any difference? Then slot in a good preamp or attenuators and use no digital attenuation. What sounds better?

I have compared the two many times via my Truth and direct. Most of the time the Truth sounds a bit better - but in some cases like the Metrum, JK Saber (actually with that one it sounded a lot better) and PDX direct sounds a bit better. In others like the Playback Designs, where you can easily hear a lot of noise direct connected, the Truth is easily better. These days I tend to direct connect purely because my PDX is direct connected and when I put a new DAC in I want to do the minimum of cable switching and the difference most of the time is not great. Right now I have lent my PDX and have Truth in there all the time even with DAC's like the M51 which are meant for direct connection.

Bottom line here is you need to suck it and see.

Thanks

Bill

[kdoot]

What this article explains better then Ive ever been able to is that digital volume control results in the signal sinking into the noise floor with a loss in dynamic range.

That's extremely important for any system planning with respect to choosing volume control.

Preamps still have a valid place. Why? To put it simply they attenuate both the noise and the signal preserving the SNR and range.

Except that pre-amps do exactly the same thing. They take a signal from upstream (which contains some noise of its own) and then manipulate it in some way, adding new noise and distortion in the process.

It's very much system-dependent as to which approach is going to give you the best result. With my present DAC and amplifier, I definitely prefer the immediacy, dynamics, detail and tonal balance I get from direct connection and digital attenuation - though I have to compromise a little at very low signal levels with whether I use dither or not, or which kind of dither to use.

Other people listening to my gear might have different preferences, and certainly with other equipment the answer will sometimes be that doing analogue attenuation sounds best.

Listen when its really quiet and turn the digital volume down a fair way e.g. -50dB and listen to the different bit depths.

Do you hear any difference?

Then slot in a good preamp or attenuators and use no digital attenuation.

What sounds better?

This will be listener- and system-dependent.

Thats excellent info thanks. I wonder how loud the undithered harmonics can be ?

They can be extremely noticeable, especially on simple/sparse tracks.

I also wonder what's the difference in volume error using 16bit versus 24bit is? Can you check my working out as I will probably get this wrong:

I guess using 16bits there is loss of accuracy in the last bit (is that called the LSB?).

I think that reduces the integrity to 15 bit?

That will give 2 to the power of 15 discreet volume steps = 32768 over a dynamic range of 6dB x 15 = 90dB or 0.00274658203125dB per step, if my mental arithmetic is OK? I dont know if my hearing is up to tell the difference between 0.003dB steps. Does this mean 16bit encoding is enough for dB steps?

I don't think it's as simple as that - I suspect it's also related to the frequency characteristics of the noise. Higher frequency at same amplitude (remember these digital errors are all about amplitude) means higher power. The digital attenuation noise is most noticeable to me as high frequency "chalk", or as obviously digital harmonics which are vaguely like some of the funny sounds you hear sometimes when a mobile phone call goes wrong. Dithering smears all that out so that it's less distinct but it turns into this horrible coming-and-going white noise that I'd rather live without.

[Nada]

Bottom line here is you need to suck it and see.

So can you try manipulating bit depth and report what you hear please?

To test connect directly without a preamp and use a software player to put out 16bit/24bit/32bit (hopefully your transport-DAC will accept those bit depths).

Listen when its really quiet and turn the digital volume down a fair way e.g. -50dB and listen to the different bit depths.

Do you hear any difference?

Except that pre-amps do exactly the same thing.

Are you sure? Digital volume deteriorates the SNR and shrinks the dynamic range out of the DAC.

That means the source is inherently compromised.

A poorly designed preamp can mess things up and being single is better then suffering a bad marriage.

While a poorly designed preamp can inject a heap of noise thats very different to messing up the SNR from source.

This theory predicts the best SNR and dynamic range will be produced using no digital volume attenuation by having a transparent preamp doing analogue attenuation duties with vanishing low noise and distortion.

Thats not what I want to believe. Digital volume control is smooth, convenient and free.

[sir sanders zingmore]

What if the noise is inaudible?

[bhobba]

What if the noise is inaudible?

You still need to suck it and see eg the Tranquility DAC is as quite as they come yet sounds much better with the Truth. For that DAC there is two reasons - one it is 16 bit so you bit drop and secondly while it has a very low output impedance the expensive coupling cap it uses is of a low value to keep costs down so if you use too low an input impedance the bass is rolled off - the truth has an immeasurably high input impedance. I hope it illustrates the reasons a pre-amp may or may not sound better are many an varied so no hard and fast rules can be given.

Thanks

Bill

Edited July 28, 2012 by bhobba

[Arg]

Except that pre-amps do exactly the same thing. They take a signal from upstream (which contains some noise of its own) and then manipulate it in some way, adding new noise and distortion in the process....

That's right. The real myth is that digital volume controls degrade the sound in a way that analogue volume controls don't. Oh yes they do. [Edit: I mean, analogue and digital both degrade SNR as the volume is turned down]

Edited July 29, 2012 by Arg

[Addicted to music]

That's right. The real myth is that digital volume controls degrade the sound in a way that analogue volume controls don't. Oh yes they do. [Edit: I mean, analogue and digital both degrade SNR as the volume is turned down]

+1 agreed,

but my experience with preamps with traditional volume pots was such an eye opener that if you have experienced what I have experimented, a digital volume control at any bit lenght will out weight any benefits that is gained against any traditional pot that includes Alps, Bourns, etc and yes I have experimented with plastic film. So dont kid yourself with traditional volume control. The main 2 points to learn here when deciding on the 2 topolgy is wear and tear and the main one is consistancy. A digital control will outlast any expensive electromechanical contact point control.

Edited July 29, 2012 by pchan

[Nada]

electromechanical contact point control.

what about those discreet resistor units that rotate?

[ehtcom]

what about those discreet resistor units that rotate?

Hi Nada.

I'll measure some pots for noise and distortion tomorrow if you like.

I have at hand.

Gold Point stepped attenuator.

Alps Blue Velvet.

Cosmos sealed carbon.

Tyco carbon

$3 no-name carbon

Cheers, Earle

[Nada]

Hi Nada.

I'll measure some pots for noise and distortion tomorrow if you like.

I have at hand.

Gold Point stepped attenuator.

Alps Blue Velvet.

Cosmos sealed carbon.

Tyco carbon

$3 no-name carbon

Cheers, Earle

That would be great. Thanks very much.

[Addicted to music]

Hi Nada.

I'll measure some pots for noise and distortion tomorrow if you like.

I have at hand.

Gold Point stepped attenuator.

Alps Blue Velvet.

Cosmos sealed carbon.

Tyco carbon

$3 no-name carbon

Cheers, Earle

Measuring them out of the circuit is pointless exercise. they all are going to appear and measure well from the start. Because 2Vp-p audio is so close to DC it is the contact resistance that matters so the matterial used is critical for stability. It isnt till so many turns that the contact resistance increases or the materials used wear and the SQ deteriorated at an expedential rate.

Its the consistancy that matters and all mechanical contact points will wear.

The best experiment is to have 2 pots, put one in the circuit and use it for 3 to 6 months, then replace it with a brand new one and I can guarrantee 100% that the new one will sound better.

With digital control it is stable, and not to mention the accuracy for 2 channels that a mechanical pot will never match!

Edited July 29, 2012 by pchan

[Addicted to music]

what about those discreet resistor units that rotate?

Nada,

these are an improvement in theory but I have never got a change to try them. I was going to get a stepped resistor network that are switched via SS relays, but the long path that the signal goes through outweighs the benifits. WFS uses them in most of there volume controls. I ended up with the M51 and nothing will match the digital control in this unit. The consistancy as you increase volume from -80 to -45db is unmatched between left and right, something I have never experienced with normal pots. I can listen at very low volume with superb channel matching, something I couldnt acheive even with expensive Alps

[ehtcom]

Data Below.

Cheap interconnect, measured, then chopped in half with each attenuator being soldered into position for the test.

2v p-p input signal.

Measured at maximum volume, then -12db and -24db.

[Nada]

Superb work there Earle. Thanks very much. The Goldpoint performs well as expected but its the no-name that intrigues me.

Dont suppose you might be interested in doing the same with a digital volume control to demonstrate what happens to the Signal to Noise level with digital attenuation?

[davewantsmoore]

This theory predicts the best SNR and dynamic range will be produced using no digital volume attenuation by having a transparent preamp doing analogue attenuation duties

This is my thinking also

[Addicted to music]

Data Below.

Cheap interconnect, measured, then chopped in half with each attenuator being soldered into position for the test.

2v p-p input signal.

Measured at maximum volume, then -12db and -24db.

@ what settings at the pot did you take these measurements????

2ndly, it be interesting to place the Goldpoint and the no name brand and give them 3-6 months use, then take the measurements at the same point, or the point at where the contacts are most used.

Another point to consider is that as you solder in and out of the circuit the heat transfer to the resistive wafer can alter the properties because most are carbon and some are conductive plastic, Like I said in a previous post its usuage and consistancy verses time. These are going to measure great regardless of what test you throw at them.

FWIW:

the 1st thing that you are going to notice in pot wear is a drop in the low end, a reduction in bass this is because audio is so close to DC that any increase in contact resistance will effect this region the most. The other thing you are going to realise is that the highs are also effected and the cymbles are less defined. In a nut shell the straight curve starts to look like a sad curve. The SQ has mid requency domination and the worst thing about it is that the deterioration is not gradual noticeably to the listener and it isnt until you place 2 preamps side by side and do a A/B comparision! I know because I built 2 identical preamps and found that the newer one I built destroyed the 1st one in SQ that was completed 6mths earlier! The culprit: volume pot !!!

Edited July 30, 2012 by pchan

[ehtcom]

@ what settings at the pot did you take these measurements????

2ndly, it be interesting to place the Goldpoint and the no name brand and give them 3-6 months use, then take the measurements at the same point, or the point at where the contacts are most used.

Another point to consider is that as you solder in and out of the circuit the heat transfer to the resistive wafer can alter the properties because most are carbon and some are conductive plastic, Like I said in a previous post its usuage and consistancy verses time. These are going to measure great regardless of what test you throw at them.

FWIW:

the 1st thing that you are going to notice in pot wear is a drop in the low end, a reduction in bass this is because audio is so close to DC that any increase in contact resistance will effect this region the most. The other thing you are going to realise is that the highs are also effected and the cymbles are less defined. In a nut shell the straight curve starts to look like a sad curve. The SQ has mid requency domination and the worst thing about it is that the deterioration is not gradual noticeably to the listener and it isnt until you place 2 preamps side by side and do a A/B comparision! I know because I built 2 identical preamps and found that the newer one I built destroyed the 1st one in SQ that was completed 6mths earlier! The culprit: volume pot !!!

Hi pchan.

1. The settings are quoted in the table.

2. Won't make much different 6 months work. Most users, myself included will only adjust the volume from time to time. Though I do have a 3~4yo Troubadour with an Alps volume control in it. If I get some spare time I'll remove it and test.

Usage and consistency will be fine for a while. It will not last like a digital volume control, but that is the price you pay for sound quality.

3. Heat isn't an issue. I'm a very speedy solderer

I'm not sure how you came to your conclusions in you "FWIW" para.

As the volume reduces, it gets closer to zero volts, not dc. It will still be the music (ac) signal, just at a reduced level p-p.

Why would the highs be affected at just say 16khz to 20 khz. A pot can transfer frequencies up to and beyond 1mhgz. Anyway the contact resistance doesn't come into it. At low volume, say a 0.05vac signal, driving a typical small signal valve. The input grid resistance will be well over 1meg ohm, and as high as 2 in some cases. The voltage drop across any contact resistance will be so insignificant I doubt it could be measured, maybe a few pico-volts. The straight "curve", will look, like a straight line 20hz to 20khz.

Cheers, Earle.

[gainphile]

Excellent effort there Earle !!

[alcarp]

Using the reasoning in the opening post of this thread will I get a better result in using less digital attenuation of the NAD M51 and using passive analog attenuators at input to the power amp?

[Lil Caesar]

....Potentially yes BUT only when listening at less that say -50Db - and that's the sticking point - you're unlikely to every listen at such a low level anyway.