Reducing DAC volume doesn't necessarily decrease bit-depth of audio data.

[Trell]

Reducing DAC volume (by say 3dB) is beneficial as it lowers distortion and allows for intersample overs.

Some DACs have digital headroom to allow for intersample overs, like the RME ADI-2 DAC FS with 2.5 dB on main output.

 

[solderdude]

Yep, some do, but not all.
On top of that most DACs have increasing distortion near FSD so -3dB or so will lower distortion.
Here too there are exceptions to that 'rule'. The D10S is such an exception.

IMO at these levels it is all moot... still... reducing DAC output digitally a few dB will never hurt.

 

[Deleted member 58865]

Reducing DAC volume (by say 3dB) is beneficial as it lowers distortion and allows for intersample overs.

Why specifically 3dB attenuation in DAC volume? What will it do to Topping D10s?

 

[Trell]

Why specifically 3dB attenuation in DAC volume? What will it do to Topping D10s?

Help explain intersample overs, please?

The Benchmark DAC3 webpages explain the concept of intersample overs and how they can lead to distortion when upsampling from 44.1 or 48 kHz to higher rates. I get that, and I think I understand the math behind it. What I’m less clear about is the solution. Benchmark talks of designing in...

www.audiosciencereview.com

A post from RME in that thread:

Help explain intersample overs, please?

Again not necessarily. But there's some truth to that. One counter example is using linear phase in AD and minimum phase in DA there is chance that this may still happen. Also if the filter of the AD is steeper than the DA it also may happen. So then I’m back to my original question... Let’s...

www.audiosciencereview.com

 

[RayDunzl]

24 bits file - before and after 60dB digital attenuation

Below:

1/48th smoothing
In-room, 40dB analog vs 40dB digital attenuation

One, the MiniDSP is wide open and the preamp is attenuated
Other, the digits are attenuated at MiniDSP and the preamp is wide open.

No "room correction" applied at the MiniDSP in this example else the trace would be much flatter.

40dB is the limit of the range of the preamp volume to "wide open" from the initially attenuated playback level chosen.

Other test measures are equally similar, phase, distortion, impulse, ETC, step, group delay.

 

[Deleted member 58865]

On top of that most DACs have increasing distortion near FSD so -3dB or so will lower distortion.
Here too there are exceptions to that 'rule'. The D10S is such an exception.

What is FSD? Why do most DACs have distortion near FSD? Why is D10s an exception?

If I search for FSD, I get tesla's full self driving.

 

[solderdude]

Full Scale Deflection = max output voltage = 0dB = 2V.

 

[little-endian]

It is absolutely about noise floor. The only difference between 16 bit, 20 bit, 24bit or 32 bit (or even 8 bit) is the level of the (quantisation) noise floor.

Mind you the same is also true for video. I hate it when even in official white papers they postulate that "8 bit quantization quals 256 shades" non-sense which always implies the absence of dither (which, to be fair, idiotically is very often omitted with video, leading to ugly banding artifacts).

 

[Tangband]

Less noise from the DAC. Not less noise from the amp which is the limiting factor here not thermal noise of a 2.5k resistor in worst case scenario (-6dB setting).
Reducing DAC volume (by say 3dB) is beneficial as it lowers distortion and allows for intersample overs.

So technically, -3dB digital + reducing gain of the following amp will give best results.
As the latter is not possible -3dB digital + passive attenuation will give you the most 'technical optimal' performance.

Hmm…. Maybe this is also a case of ” try and listen ” what sounds best ? A passive attenuation before a poweramp can possible make the dynamics slightly worse in some cases, and to much digital attenuation before the dac might worsening the sound also, even though it seems that modern 24 bit delta/sigma dacs are much better and more linear below -100 dB than ancient multibit dacs from 1990.

 

[antcollinet]

On top of that most DACs have increasing distortion near FSD

New info for me. What is the reason for that?

 

[solderdude]

A lot of DACs do this. Probably because at max. output level the used circuits come near power supply rails where non linearities are around the corner.
Not all DACs do this but a lot of them do.
Still... so far below any audible levels that is is all moot but technically most DACs perform best a few dB before max. values are reached. With music only short peaks reach these levels anyway.

A passive attenuation before a poweramp can possible make the dynamics slightly worse in some cases,

I don't think so. It probably just cannot reach loud enough levels anymore or the psychological effect of having the volume control up high yet no impressive sound is more likely to be the culprit.

 

[Philbo King]

Why do you assume that DAC has to use 24-bit fixed point numbers for analog conversion? Is there any hidden non-man-made natural law that enforces this?

What if I want to use 32-bit integers for analog conversion?

The noise floor of a physically realizable DAC analog output circuit is around -120 dBFS unless you use cryogenic cooling. Thats about 20 bits. Physics has limits...
Lets say your system is set so a 0dBFS peak generates 100 dBSPL and uses that DAC with a -120dB noise floor. Even with 20 bits, the smallest measurable signal would produce a sound level of -20 dBSPL, which is the level in a decent anechoic chamber. The blood running through your veins would be louder than that (I know, I used to work in one occasionally).
At some point, worrying about miniscule inaudible signals becomes obsession...

 

[Tangband]

A lot of DACs do this. Probably because at max. output level the used circuits come near power supply rails where non linearities are around the corner.
Not all DACs do this but a lot of them do.
Still... so far below any audible levels that is is all moot but technically most DACs perform best a few dB before max. values are reached. With music only short peaks reach these levels anyway.


I don't think so. It probably just cannot reach loud enough levels anymore or the psychological effect of having the volume control up high yet no impressive sound is more likely to be the culprit.

I have ordered some rca attenuators to have between my Rega dac and my HYPEX 252 amplifier . Im using a digital source for volumeregulation right now. In normal use , I have to digitaly attenuate the signal about -30 dB . So I’m 30 dB from 0 level in my dac.

I will try a 20 dB rca attenuation between dac and poweramp and see if the sound gets better or worse . Late next weekend I will report about my findings in this thread .

 

[Tangband]

24 bits file - before and after 60dB digital attenuation

Below:

1/48th smoothing
In-room, 40dB analog vs 40dB digital attenuation

One, the MiniDSP is wide open and the preamp is attenuated
Other, the digits are attenuated at MiniDSP and the preamp is wide open.

No "room correction" applied at the MiniDSP in this example else the trace would be much flatter.

40dB is the limit of the range of the preamp volume to "wide open" from the initially attenuated playback level chosen.

View attachment 276833

Other test measures are equally similar, phase, distortion, impulse, ETC, step, group delay.

Interesting .
Can you measure the same with a 16 bit file ? - the most common audio format.

In that case , what happens using digital attenuation -60 dB without dithering ?
Or -30 dB ?

Many streamers dont have dithering .

 

[solderdude]

It will be the same when attenuation is done digitally in 32 bit when using a 24bit resolution DAC.
The issue with 16 bit and attenuation occurs when the DAC is only 16 bit resolution or the calculations are in 16 bit.
With 16 bit level DACs you can use dither to increase dynamic range and trading in one type of noise for shaped noise.

 

[antcollinet]

I have ordered some rca attenuators to have between my Rega dac and my HYPEX 252 amplifier . Im using a digital source for volumeregulation right now. In normal use , I have to digitaly attenuate the signal about -30 dB . So I’m 30 dB from 0 level in my dac.

I will try a 20 dB rca attenuation between dac and poweramp and see if the sound gets better or worse . Late next weekend I will report about my findings in this thread .

My guess is you won't hear a difference (at least if you test blind - which might be tricky )

 

[RayDunzl]

Can you measure the same with a 16 bit file ? - the most common audio format.

An original (WAV from CD) and -60dB with 16bits

Zoomed in to show sample values

There can be visual differences in the attenuated wave - near zero crossing here.

Using dBfs display of the short segment just above:

 

[solderdude]

One should realize that this is only the calculated waveform. It is about calculations only in 32bit (or 64 bit) digital format for the attenuation.
It will not look the same in the real analog world as there will be noise added by the electronics. This is not seen on these plots.

With a 16 bit DAC you will clearly see 'steps' and quantization errors. With dither you will see shaped noise being added to the attenuated signal.
With a 24 bit or 32 bit converter (not just the input format) there will be background noise.
At least when you would actually measure the analog out from a DAC.

 

[antcollinet]

There can be visual differences in the attenuated wave - near zero crossing here.

But that is just the sampled value - not the reconstructed waveform - which (assuming dither is used) will be less dissimilar (or identical), except for the added (and shaped) quantisation noise.

EDIT : Ninjad by @solderdude

 

[Tangband]

I have ordered some rca attenuators to have between my Rega dac and my HYPEX 252 amplifier . Im using a digital source for volumeregulation right now. In normal use , I have to digitaly attenuate the signal about -30 dB . So I’m 30 dB from 0 level in my dac.

I will try a 20 dB rca attenuation between dac and poweramp and see if the sound gets better or worse . Late next weekend I will report about my findings in this thread .

Now I have my attenuators .

Using -20 dB before the poweramp there is nothing worse in the sound . Maybe a little bit of an improvement with Yamaha wxc50 as digital volumeregulation.

With -20 dB attenuating I can also use my Yamaha as a streamer turning the digital preamp off , with only one rather loud volume . But its a rather interesting comparison.

Its clear that the sound this way is better , and that the digital volume control with the Yamaha wxc50 as preamp, is not very good. I can hear the sample rate conversion to 48 KHz .

Conclusion : I will buy a WiiM pro but maybe ” only ” the pro version, and use it as a digital preamp with volume regulation avoiding sample rate conversion.