No Noise-Floor in a Digital System?
Old 17th June 2008
  #31
Lives for gear
 

Quote:
Originally Posted by DarkEcho View Post
I have become totally lost in the sea of technical jargon..

Can someone just answer this question in terms a 3rd grader would understand?

What will happen (sound quality/noise-wise) if I:
1. Track at -40dBFS and bring it up to -6dBFS versus doing the initial tracking at -6dBFS?

2. Take a track recorded at -6dBFS, lower it down to -40dBFS, bounce it, re-import it, and then bring it back up to -6dBFS?

Will there be any difference in the outcome of 1. versus 2.?
I am assuming that when you say track at -40dB you mean reduce the level after the ADC by 40dB?

It will depend on whether there is a 24bit path anywhere (TDM or PowerCore) and whether it is dithered. It will also depend on how you reduce the level before the recording (at 24bits) and whether the gain reducer provides dither.

The worst that should happen (even if there are 24bit paths) is that the noise will increase - potentially to -100dB from -140dB or so, all else remaining the same.

The worst that could happen (if there are 24bit paths that are not dithered - i.e. your recording file) is that you will have signal dependent distortion at -100dB - which may reduce your sound quality.

------

Reducing level by 40dB before the ADC in the analogue domain is a different matter entirely.

In this case the ADC noise (maybe at around -110dB) will increase to around -70dB when you bring it back up in the digital domain - which is quite a bit and not advisable.

But having said that, the actual signal quality in terms of distortion will probably be unaffected with modern ADCs.

And as a very final twist - if you have lots of ADC noise and your W/S has undithered signals (either at the recording file, in the mixer or plugs) - and you don't care too much about noise in a loud (rock) production - it is possible that the ADC noise will dither those signals and it could actually end up sounding better, despite the extra noise.

I hope this helps.
Old 17th June 2008
  #32
Lives for gear
 

Thread Starter
Quote:
Originally Posted by Ethan Winer View Post
This is so easy to do, why don't you just try it yourself?

I doubt that I'd be able to hear (the/a) difference if there was one. My ears are dumb and my gear isnt the highest of quality.

Ethan! I thought you were always the advocate for not using your ears?

If there is a plugin or something I can use to mathematically measure the difference, I would be happy to do it myself! I figured it would be a quick answer.
Old 17th June 2008
  #33
Lives for gear
 

Thread Starter
Quote:
Originally Posted by Paul Frindle View Post
I am assuming that when you say track at -40dB you mean reduce the level after the ADC by 40dB?
I am saying that you setup 1 mic that splits off to two preamps and into the DAW on two different tracks (both at unity).

You turn the gain down on one of the preamps so that one of the tracks in the DAW reads -40dBFS while the other one reads -6dBFS while they record the same source sound from the same mic.

now you stop recording and you have two tracks of the same thing, but one is a lot quieter. So you slap a gain plug on the quiet one and bring it up so that its equal to the -6dBFS signal.

I am wondering what sort of quality difference there is between these two tracks without dither anywhere in the DAW.

I'd do it but I dont have an appropriate splitter cable and I am not sure if the difference would be noticeable. I just want to know- will the one track recorded at -40dBFS (then boosted by 34dBFS) not sound as good as the one recorded at -6dBFS?

If there is inherent noise in the preamp or ADC, those will get boosted when you gain up the quiet signal in the DAW, right? Would this be the noise floor?
Old 17th June 2008
  #34
Lives for gear
 

If you split the same signal off the DAW is receiving the same signal at two different levels...

First of all, I guess we should assume that you're using a theoretically-ideal microphone preamp, because if you have the gain on on one channel of a preamp down 40 dB from the next one with an identical signal present at its input, my guess that that the difference at that stage would be more noticeable than anything that happens when the signal is converted to digital...

So assuming that your'e feeding the DAW two identical signals and one is just 34 dB quieter than the other, and then bring the quieter signal back up 34 dB digitally so the peaks "match"...

What would happen would depend on the dynamic range of the source you were recording, as well as the dynamic range of the A/D converters. In a nutshell, the two signals should be identical, but the noise floor on the second signal would be 34 dB higher than on the first, since the noise floor would be at the same level for both signals. Say we were working with a good 24-bit converter, and the noise floor was at 120 dB below full scale...then in the one signal that was recorded peaking at -6dB, the noise floor would still be at -120 dB, where on the second signal, the noise floor would be up at -84 dB. Other than that, the signal should be identical...no more or less "accurate" in one instance than the other.

In reality, the dynamic range of the source may well be small enough for there to be no noticeable difference...if it was high enough above the noise floor in both instances, that is. Say we were recording a signal with a dynamic range of, say, 50 dB...even peaking at -40 dBFS, the quietest part of the signal you're recording will be at -90 dBFS, or 30 dB above the noise floor...so both signals should still be pretty much identical, even when the level of the second signal is brought back up.

If the dynamic range of the source is large enough, though, that on the track recorded peaking at -40 dBFS the quieter parts of the signal are down low enough that you can hear the noise floor, that will be an issue when you bring them back up as well.
Old 17th June 2008
  #35
Gear Guru
 
Ethan Winer's Avatar
 

Quote:
Originally Posted by DarkEcho View Post
I doubt that I'd be able to hear (the/a) difference if there was one. My ears are dumb and my gear isnt the highest of quality.
If you do this at 16 bits your ears will easily hear the degradation!
Old 17th June 2008
  #36
Lives for gear
 

Quote:
Originally Posted by DarkEcho View Post
I am saying that you setup 1 mic that splits off to two preamps and into the DAW on two different tracks (both at unity).

You turn the gain down on one of the preamps so that one of the tracks in the DAW reads -40dBFS while the other one reads -6dBFS while they record the same source sound from the same mic.

now you stop recording and you have two tracks of the same thing, but one is a lot quieter. So you slap a gain plug on the quiet one and bring it up so that its equal to the -6dBFS signal.

I am wondering what sort of quality difference there is between these two tracks without dither anywhere in the DAW.

I'd do it but I dont have an appropriate splitter cable and I am not sure if the difference would be noticeable. I just want to know- will the one track recorded at -40dBFS (then boosted by 34dBFS) not sound as good as the one recorded at -6dBFS?

If there is inherent noise in the preamp or ADC, those will get boosted when you gain up the quiet signal in the DAW, right? Would this be the noise floor?
Ok - it's this part of my previous answer:

Reducing level by 40dB before the ADC in the analogue domain is a different matter entirely.

In this case the ADC noise (maybe at around -110dB) will increase to around -70dB when you bring it back up in the digital domain - which is quite a bit and not advisable.

But having said that, the actual signal quality in terms of distortion will probably be unaffected with modern ADCs.

And as a very final twist - if you have lots of ADC noise and your W/S has undithered signals (either at the recording file, in the mixer or plugs) - and you don't care too much about noise in a loud (rock) production - it is possible that the ADC noise will dither those signals and it could actually end up sounding better, despite the extra noise.

Oh and Ethan is right - you may even generate enough noise from all you ADCs to produce a dithered 16bit output without adding the dither internally - provided that you don't fade the master at the end of the track :-).
Old 18th June 2008
  #37
Lives for gear
 

Quote:
Over the years I have had terrible differences of opinion with people who were horrified by me constantly injecting uncertainty to mathematical processes. Their opposition to it was almost religious in that they were deeply offended that I was removing what they considered to be it's fundamental value - certainty as a philosophically comfortable property that they themselves had become disciples of. It ws frustrating sometimes.
I can well understand. However my take on it is that such a mathematical system without the noise isn't pure - it is wrong. The easy hard edged traditional mathematics has all sorts of advantages, and indeed there is a certain comfort zone operating with systems that are totally deterministic. But it doesn't mean they are fit for purpose.

Recently we had a vigourous discussion in an adjacent thread about the effective bit depth of tape. Same deal in many ways. People are not comfortable with the idea of a fraction of a bit of information as a useful metric. Sort of like people not being comfortable with negative numbers, irrationals, or worst of all - complex numbers. A mathematical system where you carry entropy about with you along with the information is just another tool. Albeit one that is a little weird for some people to contemplate.
Old 20th June 2008
  #38
Lives for gear
 

Quote:
Originally Posted by Francis Vaughan View Post
I can well understand. However my take on it is that such a mathematical system without the noise isn't pure - it is wrong. The easy hard edged traditional mathematics has all sorts of advantages, and indeed there is a certain comfort zone operating with systems that are totally deterministic. But it doesn't mean they are fit for purpose.

Recently we had a vigourous discussion in an adjacent thread about the effective bit depth of tape. Same deal in many ways. People are not comfortable with the idea of a fraction of a bit of information as a useful metric. Sort of like people not being comfortable with negative numbers, irrationals, or worst of all - complex numbers. A mathematical system where you carry entropy about with you along with the information is just another tool. Albeit one that is a little weird for some people to contemplate.
The thing is that because of the way our minds work mathematics is the only method we have to evaluate things and it's hugely useful to us in all aspects of life. But we should not lose sight of what it really is - and promote selected facets of it into religious style doctrines. The biggest problem in audio is applying it in an incomplete manner and drawing false conclusions. Like all things, a balanced view and an open mind is required :-)
New Reply Submit Thread to Facebook Facebook  Submit Thread to Twitter Twitter  Submit Thread to LinkedIn LinkedIn  Submit Thread to Google+ Google+ 
 
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
faeflora / Music Computers
14
Westmalle / Remote Possibilities in Acoustic Music & Location Recording
10
crabtwins / Low End Theory
7
lukejs / So much gear, so little time!
7
RainbowStorm / So much gear, so little time!
5

Forum Jump
 
Register FAQ Search Today's Posts Mark Forums Read

SEO by vBSEO ©2011, Crawlability, Inc.