DAC Resolution Test and Don't EVER use Digital Volume Control

[mswlogo]

I have been struggling for years why my system sounded different when I bypassed my digital Preamp (which attenuated 3dB due to Speaker Calibration). I've always thought heck it's all 24bit once it gets in the preamp (similar to how SqueezeBox/Transporter does digital volume). 24-bit does NOT matter. You attenuate 6dB digitally and you WILL lose 1 bit of resolution you used to hear. Just because it's 24-bit doesn't mean you have 8bits (8 * 6dB) 48dB to play with in volume before it starts to eat into your 16bit RIP. That's NOT how things work.

It's relative to what you actually can hear, your room, your amp, your (imperfect DAC). So when your sitting down listening at normal volumes you may get 17bits (or less). Now you start attenuating digitally even as little as 6dB now you are down to 16bits. And so on.

I think many on this forum know to always run with Volume Disabled (set to Max).

But I see so many folks run straight to a DAC and an AMP with no peramp. It is really unwise if you do. And your wasting your money. Unless you listen at near 99.

I finally came up with a test to prove it.

This is my DAC resolution test.

Try running the Test with "Max" Digital Volume (all test files are VERY low volume), then try it Digitally attenuated even 6db. The test will get an audible different result showing you audibly will hear one less bit.

The key to understanding this is just because a DAC accepts 24bits does not mean it's true performance 24bit. Your ears, system, room, amp, dac etc. have a total fixed dynamic range and by using digital attenuation you are sliding the music out of that range.

http://www.meridianunplugged.com/wik...olution%20Test

[cliveb]

It's actually quite difficult to understand exactly what your test involves, but looking at your linked page, I get the impression that you are proposing this:

1. Take a file with a very low level signal (eg. -80dB).
2. Play it without digital attenuation, and increase the analogue gain until you can hear it.
3. Now apply some digital attenuation and increase the analogue gain to compensate (so the overall level is the same).
4. There will be an audible difference.

Is that what you're saying? If it is, then my response is: "hey, no sh*t, Sherlock! Now tell us something we don't already understand".

And then try and justify how this extremely unrealistic scenario relates in any way to the normal listening experience.
Presumably you tried this test with a signal at normal levels (ie. not artificially low) and failed to hear a difference? What does that tell you about the real world?

Your ears, system, room, amp, dac etc. have a total fixed dynamic range and by using digital attenuation you are sliding the music out of that range.

And when you use analogue attenuation you are sliding the music out of that range.

[mswlogo]

It's actually quite difficult to understand exactly what your test involves, but looking at your linked page, I get the impression that you are proposing this:

1. Take a file with a very low level signal (eg. -80dB).
2. Play it without digital attenuation, and increase the analogue gain until you can hear it.
3. Now apply some digital attenuation and increase the analogue gain to compensate (so the overall level is the same).
4. There will be an audible difference.

Is that what you're saying? If it is, then my response is: "hey, no sh*t, Sherlock! Now tell us something we don't already understand".

And then try and justify how this extremely unrealistic scenario relates in any way to the normal listening experience.
Presumably you tried this test with a signal at normal levels (ie. not artificially low) and failed to hear a difference? What does that tell you about the real world?


And when you use analogue attenuation you are sliding the music out of that range.

What it tells you is the limit of your DAC even in the MOST EXTREME UNREALISTIC WORLD.

Even with 6dB (1 bit of attenuation) you can EASILY hear a difference in these unrealistic test files. But if you can HEAR any difference I assure you it will effect fine level detail of a real recording too. Especially if you digitally attenuate your music down into that range and you will if you use digital attenuation.

Then if your crazy enough to use digital attenuation for volume and subract like 24dB you are just tossing some serious resolution.

You're not pushing it down into lower bits and having your amp bring it back up, if your DAC can't resolve those low level bits even in EXTREME cases. Most DACs are no where near 24bit even though they may accept 24bit format.

If DACs were perfect to 24bits then it would not be a big deal using digital attenuation. This helps show what your DAC can and CAN'T do.

I found in my system that even 3dB attenuation (due to another problem which I finally found a work around for) had a huge difference in performance.

I also found problems in another system thinking I was getting 24bit but I was only getting 16. You'd never know unless you ran a test like this.

Another person found he had a noise problem with this test. Effectively you are finding the noise floor of your system with this.

It's crude but it works.

There plenty of folks that know what they are doing but also a bunch that doing some crazy stupid stuff. This may help convince them what they doing wrong.

Let's say you measure it in this extreme case and find you get 18bits. And you normally attenuate 24dB digitally (you play at 76). Now your down to 14bits. The point is it's not the difference between 24 and 20bits which you would NOT hear. It's much lower and depends on the limits of your DAC (to START)

Removed my insult and was uncalled for. Sorry

[cliveb]

What it tells you is the limit of your DAC even in the MOST EXTREME UNREALISTIC WORLD.

And pray tell why I should care about these EXTREME UNREALISTIC circumstances?

Even with 6dB (1 bit of attenuation) you can EASILY hear a difference in these unrealistic test files.

Of course you can. So what?

But if you can HEAR any difference I assure you it will effect fine level detail of a real recording too.

Riiiight... I routinely accept the assurances of strangers, of course.

Heaven forbid that, having compared my Transporter via a preamp versus direct (using appropriate passive attenuation, of course) and discovered that the direct connection sounds better, I should trust my own experimental results.

If you are perfect then just ignore it. No need to be jerk about it.

I will admit to being pretty aggressive in my first reply, which is fairly out of character for me. That's because your opening statements included calling people like me "fools". Now you've called me a jerk, so you'll understand why I've maintained my impolite attitude.

[mps]

I have a Duet and a high-quality external DAC going into my preamp. My kids use the SB Receiver volume control and are perfectly happy with it. However, I always turn the SB volume up to 100 and use my preamp.

A few times, I noticed the system isn't really sounding very good. Invariably, it ends up being because I forgot to turn the Squeezebox volume back up to 100, which very audibly corrected the problem immediately. I can't assign this to listener bias because I didn't have a clue that was the problem.

It is certainly settled to me that the digital volume control on the SB receiver degrades sound quality. However, I believe it is outputting 16 bits, so it doesn't prove whether the digital volume control can be used with 20/24 bits. My DAC literature claims 24 bit resolution, so maybe it will work if I ever upgrade to a Touch.

[mswlogo]

I have a Duet and a high-quality external DAC going into my preamp. My kids use the SB Receiver volume control and are perfectly happy with it. However, I always turn the SB volume up to 100 and use my preamp.

A few times, I noticed the system isn't really sounding very good. Invariably, it ends up being because I forgot to turn the Squeezebox volume back up to 100, which very audibly corrected the problem immediately. I can't assign this to listener bias because I didn't have a clue that was the problem.

It is certainly settled to me that the digital volume control on the SB receiver degrades sound quality. However, I believe it is outputting 16 bits, so it doesn't prove whether the digital volume control can be used with 20/24 bits. My DAC literature claims 24 bit resolution, so maybe it will work if I ever upgrade to a Touch.

All SqueezeBoxes are 24bit you don't need a touch to get to 24bit. It gives you 96Khz.

But I doubt your DAC resolves 24bit.

What you are hearing is correct and your pushing your 16bits out of the Range of your DAC.

You want Full Scale CD (top 16bits) to stay at the Top 16bit of your DAC (where it performs the best).

[Themis]

I never use digital attenuation, because, in my system, as I listen to relatively low levels, there is an audible difference with an analog one.
I did the test several times, I prefer a good analog preamp.

Those who don't hear any difference do as they like.

[mswlogo]

And pray tell why I should care about these EXTREME UNREALISTIC circumstances?


Of course you can. So what?


Riiiight... I routinely accept the assurances of strangers, of course.

Heaven forbid that, having compared my Transporter via a preamp versus direct (using appropriate passive attenuation, of course) and discovered that the direct connection sounds better, I should trust my own experimental results.


I will admit to being pretty aggressive in my first reply, which is fairly out of character for me. That's because your opening statements included calling people like me "fools". Now you've called me a jerk, so you'll understand why I've maintained my impolite attitude.

Sorry about that. You are right. I fixed my opening post.

If you adjust the Gain on the Transporter such that you are close to 100 on the volume scale for normal listing you would be ok. But if you attenuate even as little as 3dB you are losing clarity that you CAN hear.

My test just shows that it's possible and why it's not 24bit to start. You have to start with what your DAC can do first then go from there.

[Phil Leigh]

I haven't got time to get deeply into this a the moment but:

1) if you can only hear a very very quiet sound with your ear pressed against a driver I can absolutely guarantee you can't hear it at all at 1 metre/3 feet away. The inverse square rule sees to that.

2) any sound in the 3 least significant bits is going to be totally masked by other higher level sounds >99.9% of the time. This is partially why good MP3 can be VERY hard to detect...


3) The sound simply does not collapse into a catastrophically dynamic range limited, noisy mess if the digital volume is not at 100. You have to go way lower than that... Final thought. You are listening to a track. It starts to fade out. Does the sound quality collapse or does it just get quieter. You know how that fade was achieved? ...yes you've guessed...


4) I don't care if my DAC can only resolve to 20-bits since the best ADC's in the world struggle to get 20-bits of usable signal. The bottom 4 bits are always just noise...

[mswlogo]

I haven't got time to get deeply into this a the moment but:

1) if you can only hear a very very quiet sound with your ear pressed against a driver I can absolutely guarantee you can't hear it at all at 1 metre/3 feet away. The inverse square rule sees to that.

2) any sound in the 3 least significant bits is going to be totally masked by other higher level sounds >99.9% of the time. This is partially why good MP3 can be VERY hard to detect...


3) The sound simply does not collapse into a catastrophically dynamic range limited, noisy mess if the digital volume is not at 100. You have to go way lower than that... Final thought. You are listening to a track. It starts to fade out. Does the sound quality collapse or does it just get quieter. You know how that fade was achieved? ...yes you've guessed...


4) I don't care if my DAC can only resolve to 20-bits since the best ADC's in the world struggle to get 20-bits of usable signal. The bottom 4 bits are always just noise...

RE: 1) Your trying find the LIMIT of your DAC, not your ear or what you can hear normally. It's useful to know that limit. It's useful to prove your point in #4 to yourself. It also helps verify you system is truely passing 24bit to the DAC. And in one system where I thought I was, I'm not.

RE: 2) Correct, the 4 most least signifcant bits are in audible. So why shift higher (audible) bits into the least significant bits that are inaudible and cannot be even amplified back if your DAC can't resolve that low.

RE: 3) Don't assume attenuation is only achieved shifting things down digitally. You can have soft full scale data. Read up on Compression and Expansion. Even if Fading is done purely by digital amplitude why would you want to operate always "Faded" (attenuated digitally).

RE: 4) I agree with #4. Folks keep saying "but it's 24-Bit Volume" so you won't lose anything audible (that's the mistake). NO, it's more like 20-Bit Volume (or less) and you will hear a difference. Because your DAC really runs at 20Bit-ish (at BEST with your EAR plastered to driver at full volume) and the extra 4 bits in 24bit volume are completely useless.

That's why I did the test to see what the limit of the DAC is (in the best case). If the DAC came out 24bits that means with enough amplication you can get those real low level bits that were attenuated digitally back into audible range. But if your DAC is really 20bit's then what your shifting out that was audible now can't be recovered (even by amplifying) because 24bit DACs don't resolve all 24bits.