[diamondears]

Better check this out. It shows that volume control only affects SNR not resolution.

http://youtu.be/JYjHKv2_OqQ

With modern 24-bit DACs it's moot.

I'm talking about digital players.

 

[headwhacker]

 
I'm talking about digital players.

 
How do you think digital players control volume? Unless we are thinking about different things.

 

[headwhacker]

   
I always do 100% digital so it just simplifies whatever I'm doing.
 
Just being lazy really.

 
Either way is not really an issue. I also do the same thing for similar reason. But with sensitive phones (i.e. Grados), I don't worry much decreasing the volume on my Mac to get into proper listening level thinking that I'm losing something.

 

[castleofargh]

it's not really hard. CDs with great dynamic will have maybe 65db used(usually a good deal less than that). so that's about 11bit of signal. you add 20 or 30db to ensure a background noise low enough not to ever be audible, and you end up with a reasonable need for 16bit resolution as a nice comfort margin(wow it's as if they had thought about it when they came up with redbook ^_^).
so when you output your music on the odac in 24bit(as recommended), you get some nice "extra" bits with nothing but zeros that are begging you to be used for volume control on the computer. that's the real pure lazy man way

.
 
let's say that 24bit dacs are really 20bit as a conservative estimation. that's still 4bits available before "attacking" the 16bits of a CD. so about 6*4=24db you can use on you computer, with no real danger of changing anything on the record itself.
as long as you use proper settings and know what you're doing there is no risk of degrading your music.
 
on a 16bit DAC (does that even exist anymore?) I wouldn't give you the same speech obviously.

 

[jseaber]

Software should always be 100%, otherwise you lose some digital bits. Headroom should be on the amp volume/power.

 
Every DAC we've measured performs dramatically worse at full-scale, 100% volume. Decrease your software volume slider by -0.1dB (a single notch will do) for optimal THD+N performance.
 
See Gibbs phenomenon:
http://en.wikipedia.org/wiki/Gibbs_phenomenon

 

[Medaud]

Every DAC we've measured performs dramatically worse at full-scale, 100% volume. Decrease your software volume slider by -0.1dB (a single notch will do) for optimal THD+N performance.

See Gibbs phenomenon:
http://en.wikipedia.org/wiki/Gibbs_phenomenon

Are we talking the windows volume slider here or what software exactly?

 

[jseaber]

Are we talking the windows volume slider here or what software exactly?

 
Use any digital volume slider at your disposal. The key is to avoid a full-scale signal into the DAC.

 

[blse59]

   
Use any digital volume slider at your disposal. The key is to avoid a full-scale signal into the DAC.

Wow, this is the first I've heard this. Practically every article/blog I've read regarding this issue says to keep digital volume at 100%. Thanks.

 

[Solrighal]

In the days of 16 bit DAC's it would be good practise to use software volume at 100% but since the advent of 24 bit DAC's.

 

[jseaber]

 
Wow, this is the first I've heard this. Practically every article/blog I've read regarding this issue says to keep digital volume at 100%. Thanks.

 
It was a surprise when I first noticed the measurements. After checking different units and DAC models, and conferring with other engineers, it was clear that full scale DAC performance is rarely discussed.
 
THD+N performance is 2-3x better for input just below 0dB. Keep in mind you're rarely going to hit 0dB peaks when listening to music. Even today's heavy handed recordings will only hit 0dB peaks for a few seconds throughout an entire track. And even if you do listen to a track full of 0dB peaks, 100% digital volume may yield THD+N of 0.0070% at peaks through ODAC (worst case), versus ~0.0025% at slightly reduced digital volume. It takes focused, well trained ears to detect this difference. We expect transparency at THD+N below ~0.005%. It's a rather small audible difference, yet objectively/measurably obvious.
 
We include a short blurb in the Listening Tips of JDS Labs instruction Guides:
 

For maximum signal fidelity, keep ODAC digital volume set slightly below 100% at the computer. Only control volume from your headphone amplifier.

 
Zorrofox is correct. Digital volume control is fine to use aggressively as long as the DAC has excellent dynamic range.
 

In the days of 16 bit DAC's it would be good practise to use software volume at 100% but since the advent of 24 bit DAC's.

 

 

[Medaud]

Wow, this is the first I've heard this. Practically every article/blog I've read regarding this issue says to keep digital volume at 100%. Thanks.

Also the first I heard of it. I've been running my volume at 100% up til now. Now I turned down the windows volume slider one notch...the OCD in me is acting up but I think I'll survive

 

[diamondears]

First time for me too. Thanks for that, JSeaber.

It was a surprise when I first noticed the measurements. After checking different units and DAC models, and conferring with other engineers, it was clear that full scale DAC performance is rarely discussed.

THD+N performance is 2-3x better for input just below 0dB. Keep in mind you're rarely going to hit 0dB peaks when listening to music. Even today's heavy handed recordings will only hit 0dB peaks for a few seconds throughout an entire track. And even if you do listen to a track full of 0dB peaks, 100% digital volume may yield THD+N of 0.0070% at peaks through ODAC (worst case), versus ~0.0025% at slightly reduced digital volume. It takes focused, well trained ears to detect this difference. We expect transparency at THD+N below ~0.005%. It's a rather small audible difference, yet objectively/measurably obvious.

We include a short blurb in the Listening Tips of JDS Labs instruction Guides:


Zorrofox is correct. Digital volume control is fine to use aggressively as long as the DAC has excellent dynamic range.

Thanks for that, John. First time for me too. All my DACs automatically disables the digital volume control on the source player, but never experienced anything bad on the O2s (I've got 3).

I think this has got to do with the Vrms coming out of the digital player for which the DAC input isn't designed?

 

[jseaber]

First time for me too. Thanks for that, JSeaber.
Thanks for that, John. First time for me too. All my DACs automatically disables the digital volume control on the source player, but never experienced anything bad on the O2s (I've got 3).

I think this has got to do with the Vrms coming out of the digital player for which the DAC input isn't designed?

 
Hm, I suppose you could phrase it that way. "VRMS" does not come out of a digital player, as Volts RMS is an analog unit; a digital player generates a digital signal, involving 0's and 1's as I'm sure you know. For example, the maximum signal magnitude a 24-bit digital audio player can generate = 100% volume = 0dB = 111111111111111111111111 binary
 
No amplifier is involved when measuring DAC line-output, so by definition, full scale output is that of the DAC. In the case of ODAC, its output is 2.0VRMS for input of 0dB. The test condition I've mentioned applies to any signal frequency at exactly 0dB. In other words, this is a general DAC phenomenon. It's impossible to feed a DAC more than 0dB.

 

[diamondears]

Hm, I suppose you could phrase it that way. "VRMS" does not come out of a digital player, as Volts RMS is an analog unit; a digital player generates a digital signal, involving 0's and 1's as I'm sure you know. For example, the maximum signal magnitude a 24-bit digital audio player can generate = 100% volume = 0dB = 111111111111111111111111 binary

No amplifier is involved when measuring DAC line-output, so by definition, full scale output is that of the DAC. In the case of ODAC, its output is 2.0VRMS for input of 0dB. The test condition I've mentioned applies to any signal frequency at exactly 0dB. In other words, this is a general DAC phenomenon. It's impossible to feed a DAC more than 0dB.

I see. Thanks for the time, John.

 

[castleofargh]

zorro, at last someone is saying we might not be completely crazy \o/. this is a great day for us, thank jseaber.

 
my idea about this was somehow far fetched(as always when you wrongly try to find a cause to your consequence instead of going the other way around)but I'll give it a go anyway I have no shame ^_^:
with some little jitter variations between 2 samples, it should be possible that the resulting sine would sometimes not really reach 0db when it should, and at other times try to very slightly pass 0db when obviously it shouldn't. making that tiny little part to clip. of course for that we need the signal to be at 0db and the jitter to be fluctuating in value from one sample to the next(constant lag would do nothing).
that's how I imagined it could happen. as I can't think of a reason why there should be a problem on the digital side, unless we go clip the signal with some EQ or a DSP, but it's another matter and it's easy to control.