On dac-chip volume control guidance

[Tonto]

I like to listen to music at low volume levels and would like to handle this sans a preamp. I am looking for a dac and saw the video about ESS chips have a way of keeping noise and distortion low at large attenuation levels - say over 40 db. i wonder if AKM AK@44xxEQ or other brands also accomplish this high attenuation cleanly. Have not been able to gather about this from manufacturers other than ESS.

 

[majingotan]

I seriously doubt you can hear the noise floor and distortion of a well engineered DAC at very low signal levels of -60 dB or lower. You would be definitely fine using the Windows mixer as your volume control and would probably even perform better than most preamps out there.

 

[MRC01]

You ask a good question and there is considerable discussion on this topic here if you search around.

Full scale output SINAD & SNR numbers can be misleading because most devices lose SINAD or SNR as you turn down the volume. Amir added a 50 mV SNR test to measure this and the differences are usually significant. You can do the math, but 50 mV output is -32 dB down from 2 V output, so for line level listening it is a low to moderate volume depending on your amp's gain and speaker efficiency. With typical full size headphones, 50 mV usually produces around 70 to 80 dB range, so your listening may be even quieter than this.

The results from devices measured here suggest that attenuating the signal cleanly, minimizing loss of SNR, is neither simple nor easy. It makes me question whether digital attenuation in the DAC chip is the best way to preserve SNR as you turn down the volume. You might want to read Amir's ranking of 50 mV SNR to find devices that have the cleanest output at low listening levels.

 

[RayDunzl]

I like to listen music at low volume levels and would like to handle this sans a preamp.

Visual example of Digital Attenuation:

Full digital volume vs -60dB waveform (24bit), or 1/1000th the original voltage level coming from a modern DAC.

Hint: -60dB is like going from normal level to near silence.

Where's the noise?

 

[bennetng]

ESS chips have a way of keeping noise and distortion low at large attenuation levels

https://www.audiosciencereview.com/forum/index.php?threads/is-a-preamp-needed.10154/post-277604
https://www.audiosciencereview.com/forum/index.php?threads/is-a-preamp-needed.10154/post-278701

 

[Tonto]

Yes i think the ESS video from 2011 said you could get a quieter/less distortion volume control on extremely quiet preamps. In the video he remarked the noise floor of ESS DAC chips (not sure which) was -135 dB. Does anybody know what AK449xEQ noise floor would be..i dont see in the spec sheet. Being a 32 bit dac i assume losing 10 bits would preserve the accuracy for low noise/distortion.

 

[HammerSandwich]

Where's the noise?

Hiding. (It's probably frightened of Divine.)

As you'd hope, Audacity internally uses floating-point math, so your operations truly have lost nothing. But we all know that 60dB of attenuation will increase the relative noise level. How do we find it? By removing the floats.

I imported & trimmed a 16-bit CD sample. Exported it to 16-bit FLAC. After applying 30dB attenuation twice (because Audacity won't do 60dB in one pass), I exported that version with the same settings and closed Audacity, just to be sure. Next, I restarted Audacity & imported the 2 samples. Finally, the attenuated track was amplified by 30dB, twice. Can you tell them apart now?

 

[RayDunzl]

As you'd hope, Audacity internally uses floating-point math, so your operations truly have lost nothing.

If you specify the file format for 32bit float, yes, for 16 and 24bit PCM, I think no, from past experiments. I don't think the export/import is necessary when Audacity is set for 16 or 24 bit PCM.

Example:
Second track has gone -60dB and +60dB. No export/import, and no dither. Now it has flat tops and stair-steps

Can you tell them apart now?

Sure can, but...

Whose DAC uses 16 bits for volume control?

---

Below, showing the difference between a 16bit source, using 24bit integer volume control, and attenuating/amplifying by 60dB, results in a difference (noise?) between the 16bit input and 24bit output at least 80dB down.



My dac uses 32 bit, don't know if it floats it or not, rather irrelcevant, I think.

Let's see:

"a 32-bit dithered volume control for digital inputs"

Ok, good enough.

 

[pozz]

Yes i think the ESS video from 2011 said you could get a quieter/less distortion volume control on extremely quiet preamps. In the video he remarked the noise floor of ESS DAC chips (not sure which) was -135 dB. Does anybody know what AK449xEQ noise floor would be..i dont see in the spec sheet. Being a 32 bit dac i assume losing 10 bits would preserve the accuracy for low noise/distortion.

If I'm right about the one you're referring to, I've made a thread. That presentation on volume controls is misleading at best.

https://www.audiosciencereview.com/forum/index.php?threads/two-old-ess-talks.9216/

I'm pretty sure another member highlighted math errors in the slideshow. Can't find that thread though.

 

[Willem]

I use an RME ADI-2 as a preamp and with its auto reference level systen I cannot hear any noise at low levels. The sound quality is much better than with my previous admittedly antequated Quad 33 preamp.

 

[bennetng]

I'm pretty sure another member highlighted math errors in the slideshow.

Thanks. This one:
https://www.audiosciencereview.com/...depth-should-i-set-my-dac-to.8956/post-255104

 

[AnalogSteph]

I use an RME ADI-2 as a preamp and with its auto reference level systen I cannot hear any noise at low levels. The sound quality is much better than with my previous admittedly antequated Quad 33 preamp.

Antiquated is the right word indeed - the 33 first debuted in 1967 and really should have been retired by about 1978/79 instead of being made until 1982 in typical UK fashion. It features a grand total of 4 transistors per channel (6 when using the phono preamp), with poor gain staging. It was basically outdated 40 years ago.

The contrast to the ADI-2 DAC could hardly be any greater. Mind you, that is pretty much the S-Class of DACs, with almost an entire preamp already built-in.

I do think that it is quite possible to make a preamp that is not sonically objectionable using just mid-1970s technology, e.g. based on the Radford ZD22 circuitry with maybe a pinch of late-'70s Grundig designs thrown in. It feels sort of pointless though.

 

[March Audio]

See here for a direct comparison of Roon digital volume V an analogue pot.

Spot the difference.

https://www.audiosciencereview.com/...-dig-or-not-to-dig-that-is-the-question.5951/

 

[March Audio]

@amirm can you do some testing on analogue V digital volume? I can send through the pot shown above if you don't have time to knock one up.

 

[JohnYang1997]

Noise can easily be audible with sensitive iems. Normally you control the volume and noise together with pot. But if you use digital control the output noise will be more or less constant. That's how you lose "bit" and SNR. How many bit of the digital volume control doesn't matter at all, but higher than 20bit at least. It's the dac's output noise that limits the performance.

 

[bennetng]

Noise can easily be audible with sensitive iems. Normally you control the volume and noise together with pot. But if you use digital control the output noise will be more or less constant. That's how you lose "bit" and SNR. How many bit of the digital volume control doesn't matter at all, but higher than 20bit at least. It's the dac's output noise that limits the performance.

The fact is if you can hear that noise with 24-bit dithered silence without hooking up a preamp, then adding a preamp is most likely beneficial. This can be easily done with a ABX test with a pure digital silence file and a dithered silence file.

The whole thing has nothing to do with digital at all.

 

[JohnYang1997]

The fact is if you can hear that noise with 24-bit dithered silence without hooking up a preamp, then adding a preamp is most likely beneficial. This can be easily done with a ABX test with a pure digital silence file and a dithered silence file.

The whole thing has nothing to do with digital at all.

It's not inherent issue with digital but replacing analogue attenuation with digital attenuation.

 

[bennetng]

It's not inherent issue with digital but replacing analogue attenuation with digital attenuation.

Actually this SNR issue can be happen with a completely analog chain as well, for example when listen to the output of a tape machine, analog mixer and so on. If the preamp has lower noise than the source then adding a preamp could be beneficial.

 

[Tonto]

Noise can easily be audible with sensitive iems. Normally you control the volume and noise together with pot. But if you use digital control the output noise will be more or less constant. That's how you lose "bit" and SNR. How many bit of the digital volume control doesn't matter at all, but higher than 20bit at least. It's the dac's output noise that limits the performance.

Yes that’s why the video said some ESS (older chip?) dac noise floor was -135 dB. I was wondering what it was on newer ESS chips and also recent AKM dac chips - not finding this in spec sheets.

 

[JohnYang1997]

It's the dac's output noise that limits the performance.

Yes. But with external analogue volume control the noise decreases as the volume being attenuated. So the SNR is preserved.