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

[antcollinet]

Just for the fun of it I did a loopback of the E-MU to see,first with it's crappy analog pot from the headphone amp and then just attenuating it by windows mixer (all ASIO).

View attachment 276570

at -4db


View attachment 276571

at -30db crappy pot


View attachment 276572

at -30 by windows mixer.

Fell free to comment,suggest other measurements,etc.

Windows mixer - harmonics dissappearing into the noise floor as expected.

But now I am confused regarding your test setup. What is the E-MU. What is the source? Into what? Which output you are measuring in both sets of tests? What are you measureing it with? And quesiton above - how are you accessing the ESS volume control?

 

[Sokel]

Windows mixer - harmonics dissappearing into the noise floor as expected.

But now I am confused regarding your test setup. What is the E-MU. What is the source? Into what? Which output you are measuring in both sets of tests? What are you measureing it with? And quesiton above - how are you accessing the ESS volume control?

Ok,my bad,I should describe.

Test rig in the last one is a loopback:

Pc > E-MU dac >E-MU ADC > PC.

In the first measurements:

PC > Khadas DAC > E-MU ADC >PC.

No isolators or special test conditions,the opposite,I like to test worst case scenarios.

Simple really but we should mention the test rig.

(E-MU is a 13 yo interface widely used in diyAudio,modified or not,is got some nice AKM ADC in it).

Edit:as for Khadas control:

IAN CANADA ESS SABRE 90X8 controller with display - Audiophonics

www.audiophonics.fr

 

[Sokel]

In well-implemented DAC digital volume control measurements should look like that:

Topping D90SE Review (Balanced DAC)

Merry Xmas Amir What is type of USB port at the back of this Topping D90SE DAC. Thanks. Type B. Topping includes the cable with the unit. Thanks for reaching out. Much appreciated. Which would be the preferred or better input to use on this Topping D90SE DAC? USB or Coax?

www.audiosciencereview.com

The harmonics are gone in the noise.

I just see the post,I believe that what is presented there is gain,not VC.
But we can easily see such tests lately by Amir in SINAD vs measured level measurements.

 

[Deleted member 58865]

My conclusion is that you are not supposed to reduce DAC volume too much. Physics limits DAC bit-depth to only 22~23 bits at the output.

Thus, it is a good idea to use a volume pot or a stepped attenuator for big gain changes and DAC volume for small adjustments.

 

[antcollinet]

My conclusion is that you are not supposed to reduce DAC volume too much. Physics limits DAC bit-depth to only 22~23 bits at the output.

Thus, it is a good idea to use a volume pot or a stepped attenuator for big gain changes and DAC volume for small adjustments.

wrong way of looking at it.

If you have only 20bit DR, your noise floor is down at -120dB

So lets say you turn up your amp so that your loudest music will be at 120dB (enough to damage your hearing in a very short time).

Your noise floor is now at 0dB, at the limit of human hearing and at least 20dB below the noise floor of a very quite room.

It doesn't matter how much you turn down your music digtially - you still can't hear the noise.

 

[Deleted member 58865]

wrong way of looking at it.

If you have only 20bit DR, your noise floor is down at -120dB

I was talking about bit-depth resolution.

With physical limitation of DAC bit-depth at 22 bits, if you set DAC volume to -48dB, you are left with 14 bits of resolution.

If you decrease any further, you start losing even more resolution between volume levels.

This is not about the noise floor. This is about resolution of volume levels.

With 16-bit audio, you have 65536 volume levels. As you lose more bits, you have less volume levels available.

Thus, DAC volume is best left for fine adjustments. Big gain changes should be done in a stepped attenuator or a volume pot.

 

[danadam]

With physical limitation of DAC bit-depth at 22 bits [...]
Big gain changes should be done in a stepped attenuator or a volume pot.

You assume that the attenuator won't be affected by the same physical limitations. From Johnson-Nyquist noise:

Thermal noise is present in all electrical circuits
[...]
For a 1 kΩ resistor at room temperature and a 10 kHz bandwidth, the RMS noise voltage is 400 nV.

 

[Trell]

...
Thus, DAC volume is best left for fine adjustments. Big gain changes should be done in a stepped attenuator or a volume pot.

From the several excellent posts given in this thread this is not so for modern DACs, assuming proper gain staging.

 

[Sokel]

(I would expect the harmonic distortion comes from the analoge electronics after the DAC chip. However, if that were the case, I'd expect the absolute levels of HD to be proportional to the attenuated level of the output signal (Like it is for 2nd and 4th). Increased absolute HD suggests to me that it is being added by the volume change in the digital domain - but I also don't see how this could happen)

As your post should be correct in theory,I tried to make another little test,this time in 5db segments to observe the progression of the harmonics.

I only have to say it's weird,have a look:

-5db

-10db

-15db

-20db

-25db

-30db and...

all together.

In the following link all the .mth test files are provided for everyone to have a look cause I always feel that my newbie measurements needs to be reviewed by all of our skilled friends here:

WeTransfer - Send Large Files & Share Photos Online - Up to 2GB Free

WeTransfer is the simplest way to send your files around the world

we.tl

I think that someone with proper gear and everything must repeat that,it's strange to say the least.

Edit cause a friend asked me:
For anyone not familiar, TD+N on top-right of each measurement = SINAD.

 

[Deleted member 58865]

You assume that the attenuator won't be affected by the same physical limitations. From Johnson-Nyquist noise:

So, DAC volume is a substitute for a volume pot or a stepped attenuator?

 

[Trell]

So, DAC volume is a substitute for a volume pot or a stepped attenuator?

Yes.

That doesn’t exclude use of a volume pot or attenuator after the DAC to limit max output to protect hearing or speakers. For my two Genelec 2.1 desktop setups I’ve changed the input sensitivity so that full DAC output does exactly so.

 

[Deleted member 58865]

Topping D10s has SINAD of 112dB. Digitally, it has 32-bit resolution. But, it has only 18~19 bits of actual resolution? Or?

 

[Deleted member 58865]

I think it's a bit misleading. Noise floor of Topping D10s may be -112dB, but it still has 32-bits of internal resolution.

Thus, you still have good resolution above -112dB. Correct?

After setting DAC volume to -48dB, you don't get bit depth of (112/6)-8=18.666-8=10.666.. bits. You still have bit depth of 32-8=24 bits above -112dB.

Even after setting DAC volume to -48dB, you still have more than 16-bits of resolution above -112dB. It's good to have some mathematical wiggle room.

 

[Deleted member 58865]

I guess DAC volume is just as good as a volume pot or a stepped attenuator. But, I can accidentally set DAC volume to 0dB. That's dangerous.

A volume pot has channel imbalance at low volume levels. Thus, I may firewall the DAC output with a stepped attenuator that has 3~4dB steps and use DAC volume for fine adjustments.

 

[antcollinet]

I was talking about bit-depth resolution.

With physical limitation of DAC bit-depth at 22 bits, if you set DAC volume to -48dB, you are left with 14 bits of resolution.

If you decrease any further, you start losing even more resolution between volume levels.

This is not about the noise floor. This is about resolution of volume levels.

With 16-bit audio, you have 65536 volume levels. As you lose more bits, you have less volume levels available.

Thus, DAC volume is best left for fine adjustments. Big gain changes should be done in a stepped attenuator or a volume pot.

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.

I'll post the Monty video which explains it if you insist

 

[Trell]

I guess DAC volume is just as good as a volume pot or a stepped attenuator. But, I can accidentally set DAC volume to 0dB. That's dangerous.

That's what I wrote in my reply to you and what I've done.

A volume pot has channel imbalance at low volume levels. Thus, I may firewall the DAC output with a stepped attenuator that has 3~4dB steps and use DAC volume for fine adjustments.

Sure, that can be an issue, but perhaps chose a volume pot where you can use much of the range offered by the pot? Some use a passive attenuator with a fixed attenuation instead and use the DAC volume control.

 

[Trell]

I'll post the Monty video which explains it if you insist

The one where a dude at a restaurant was fed one small morsel too much?

Edit: Since you live in UK I assumed you meant Monty Python and not that excellent Monty YouTube clip you posted.

 

[Deleted member 58865]

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.

I'll post the Monty video which explains it if you insist

I was wrong. A 32-bit DAC is useful.

Let's take Topping D10s as an example. Topping D10s has SINAD of -112dB. Let's assume the noise floor of Topping D10s is -112dB.

If I set the DAC volume to -48dB, the DAC still provides 32-8=24 bits of resolution between -112dB and -48dB.
Even after setting DAC volume to -96dB, you still have 32-16=16 bits of resolution between -112dB and -96dB.

32-bit DACs are good because they provide mathematical wiggle rooms for low-volume levels. Even at very low volume levels, you still get more than 16-bits of resolution above the DAC noise floor.

Even if application volumes demand further attenuation from the DAC volume at -48dB or -60dB, you are likely to retain 16 bits of "digital" resolution above the DAC noise floor.

32-bit DACs are not marketing gimmicks. They are actually useful.

 

[Deleted member 58865]

I conclude that DAC volume can substitute for a volume pot or a stepped attenuator if you don't care about safety.

If you care about safety, you should use a volume pot or a stepped attenuator as much as possible after DAC output.

If a stepped attenuator has 3~4dB steps, then DAC volume can be used for fine tuning.

 

[antcollinet]

I was wrong. A 32-bit DAC is useful.

Let's take Topping D10s as an example. Topping D10s has SINAD of -112dB. Let's assume the noise floor of Topping D10s is -112dB.

If I set the DAC volume to -48dB, the DAC still provides 32-8=24 bits of resolution between -112dB and -48dB.
Even after setting DAC volume to -96dB, you still have 32-16=16 bits of resolution between -112dB and -96dB.

32-bit DACs are good because they provide mathematical wiggle rooms for low-volume levels. Even at very low volume levels, you still get more than 16-bits of resolution above the DAC noise floor.

Even if application volumes demand further attenuation from the DAC volume of -48dB or -60dB, you are likely to retain 16 bits of "digital" resolution above the DAC noise floor.

32-bit DACs are not marketing gimmicks. They are actually useful.

OK - seeing as you insist :

(Hint - you are thinking about it wrong. More bits doesn't give you more "resolution" in terms of accuracy of the signal - it only gets you a lower noise floor - more dynamic range)

This is a really interesting video (I watch it regularly to remind myself how it all works - especially dithering). It is great for dispelling a number of misunderstandings about how digital audio works.