[[email protected]]

Hello,

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6 db.

if i understand correctly, the sound level will be the same as if you
never moved any levels, but the wave file going out of the master
output would now suffer degradation because you affected the bit-
depths twice. maybe the quality loss is not noticeable, but
mathematically there is a quantifiable effect.

is this correct?

[Scott Dorsey]

wrote:

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6 db.

With what console?

if i understand correctly, the sound level will be the same as if you
never moved any levels, but the wave file going out of the master
output would now suffer degradation because you affected the bit-
depths twice. maybe the quality loss is not noticeable, but
mathematically there is a quantifiable effect.

Depends on the console. The thing is, digital consoles take the
input and convert it to a very wide word, so that you can make these
changes and have plenty of room to work. It's just like the headroom
in an analogue console; you design an analogue console so all the channels
going full tilt with the faders up all the way aren't enough to clip
the buss amplifier. In the digital world, you design is so that all the
faders up all the way can't clip the summing logic, and that with them
down all the way there's as little resolution loss as possible.

A 6 dB drop from nominal on the channel strip should not be enough to cause
truncation, on a well-designed console.

Some consoles are better at this than others, and some of them use floating
point representations as intermediates. Needless to say, you can be sloppy
about gain structure and be more apt to get away with it on a Capricorn than
on an 01V.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Geoff]

wrote:
Hello,

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6 db.

if i understand correctly, the sound level will be the same as if you
never moved any levels, but the wave file going out of the master
output would now suffer degradation because you affected the bit-
depths twice. maybe the quality loss is not noticeable, but
mathematically there is a quantifiable effect.

is this correct?

Yes, though "affected the bit depths" isn't quite concise.

geoff

[Geoff]

Scott Dorsey wrote:
wrote:

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6
db.

With what console?

I think he means internal to a DAW.

geoff

[Scott Dorsey]

Geoff wrote:
Scott Dorsey wrote:
wrote:

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6
db.

With what console?

I think he means internal to a DAW.

With what software, then? A DAW is just a digital console implemented
in software, with some editing stuff.

There is software that uses 32-bit float representations and... some other
software that is not so robust.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Mike Rivers]

wrote:

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6 db.

if i understand correctly, the sound level will be the same as if you
never moved any levels, but the wave file going out of the master
output would now suffer degradation because you affected the bit-
depths twice. maybe the quality loss is not noticeable, but
mathematically there is a quantifiable effect.

Right on both counts, but depending on the word length (the internal
working word length, not just the recorded word length) you might have
to drop it by more than 6 dB to get a significant result, even as a
calculation. There used to be a classic experiment to demonstrate the
value of recording at greater than 16-bit resolution by reducing the
level by 60 dB and then cranking it back up again. That was pretty
obvious with a 16-bit recording, but just barely noticeable as a bit
more noise with a 20-bit recording.

You shouldn't let theory get in the way of practice, particularly in
(those rare) cases where practice is better than theory.

[Eeyore]

Scott Dorsey wrote:

you design an analogue console so all the channels going full tilt with the
faders up all the way aren't enough to clip the buss amplifier.

Well actually you don't !

That'll clip nicely. However that factor is indeed taken into account but has to
be compromised against noise buildup. Digital excels in this respect as long as
it has very long word lengths.

Graham

[Arny Krueger]

"Scott Dorsey" wrote in message


Depends on the console. The thing is, digital consoles
take the input and convert it to a very wide word, so that you can
make these changes and have plenty of room to work.

In real-world numbers, a minimum of 24 bits with the 144 dB dynamic range
that goes with it.

It's just like the headroom
in an analogue console; you design an analogue console so
all the channels going full tilt with the faders up all
the way aren't enough to clip the buss amplifier.

Some may, but I have never worked with a console like that.

In the digital world, you design is
so that all the faders up all the way can't clip the
summing logic,

Ditto.

and that with them
down all the way there's as little resolution loss as
possible.

All the way down on the digital consoles I've worked with turned the channel
off. However, the basic 144 dB dynamic range means that with real-world
signals, there's about as much if not more headroom as there is dynamic
range in the input signal. IOW you can have a full-scale input signal which
might have 72 dB dynamic range, attenuate it a whopping 72 dB, then amplify
it 72 dB, and end up with a signal that lost maybe a dB or two of its
original dynamic range.

A 6 dB drop from nominal on the channel strip should not
be enough to cause truncation, on a well-designed console.

Every attenuation causes loss of dynamic range in some sense, but the sense
may be irrelevant to sound quality because the signal being attenuated has
finite dynamic range because it comes from the real world.

Some consoles are better at this than others, and some of
them use floating point representations as intermediates.

True, console designers are often less than clear about exactly what happens
once the signal goes into DSP-land. I could see a console whose control
software figures out the net gain from the input to the summing point, and
tells the DSP to just apply the net, not the individual steps. Then the 6
dB attenuation followed by 6 dB gain would net out to zero, and no
attenuation would be applied.

Needless to say, you can be sloppy about gain structure
and be more apt to get away with it on a Capricorn than
on an 01V.

I see absolutely no evidence to support that contention. Based on the
documentary evidence I've reviewed from the AMS/Neve web site, they are both
basically 24 bit boxes. Differences abound - the Capricorn looks more like
an attempt to take the block-level design of an analog console and implement
it digitally. The newer Yamaha consoles implement a UI that looks something
like an analog console, but under the covers everything gets digitized and
thrown into a bank of DSPs.

[[email protected]]

Yes, a DAW, completely in the box. The ones I use are 32 float. But
then I think Protools is 40-bit fixed? Would that matter at all?

[Scott Dorsey]

Eeyore wrote:
Scott Dorsey wrote:

you design an analogue console so all the channels going full tilt with the
faders up all the way aren't enough to clip the buss amplifier.

Well actually you don't !

That'll clip nicely.

Not on my console! I can slam everything all the way up and it's all
fine. But it's true that different consoles make different decisions
about noise vs. headroom.

However that factor is indeed taken into account but has to
be compromised against noise buildup.

True, and a lot of folks have gone to active summing busses which can bring
the noise floor down at the expense of headroom.

Digital excels in this respect as long as
it has very long word lengths.

And there is no reason _not_ to have very long word lengths for internal
representations today.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Les Cargill]

wrote:

Hello,

let's say you have a 10 second wave file with "normal" levels. then
you lower it 6 db on the track fader, and raise the master fader 6 db.

if i understand correctly, the sound level will be the same as if you
never moved any levels, but the wave file going out of the master
output would now suffer degradation because you affected the bit-
depths twice. maybe the quality loss is not noticeable, but
mathematically there is a quantifiable effect.

is this correct?


If have done an experiment before that convinced me that there
is no quantifiable effect. I'll rerun it and report back.

--
Les Cargill

[Michael R. Kesti[_2_]]

Scott Dorsey wrote:

Not on my console! I can slam everything all the way up and it's all
fine. But it's true that different consoles make different decisions
about noise vs. headroom.

Can you apply perfectly correlated +14 dBu, 1 kHz sinusoids to all of your
mixer's inputs, assign all of those inputs to the the mixer's L/R buss,
unmute them, raise all of the faders to their maximum gain settings, and
not experience clipping on the L/R buss?

--
================================================== ======================
Michael Kesti | "And like, one and one don't make
| two, one and one make one."
mrkesti at hotmail dot com | - The Who, Bargain

[Eeyore]

Scott Dorsey wrote:

Eeyore wrote:
Scott Dorsey wrote:

you design an analogue console so all the channels going full tilt with the
faders up all the way aren't enough to clip the buss amplifier.

Well actually you don't !

That'll clip nicely.

Not on my console!

What console is that ?


I can slam everything all the way up and it's all fine.

I bet you can't actually !


But it's true that different consoles make different decisions
about noise vs. headroom.

No question. Some that appear very quiet may have useless headroom on the busses.


However that factor is indeed taken into account but has to
be compromised against noise buildup.

True, and a lot of folks have gone to active summing busses which can bring
the noise floor down at the expense of headroom.

No reason why they should to be honest. Did you have anyone specific in mind ?


Digital excels in this respect as long as it has very long word lengths.

And there is no reason _not_ to have very long word lengths for internal
representations today.

Given the right DSP hardware, absolutely so.

Graham

[Mike Rivers]

On Apr 27, 1:59 am, "Michael R. Kesti"
wrote:

Can you apply perfectly correlated +14 dBu, 1 kHz sinusoids to all of your
mixer's inputs, assign all of those inputs to the the mixer's L/R buss,
unmute them, raise all of the faders to their maximum gain settings, and
not experience clipping on the L/R buss?

Why would anyone do that? To break a console?

On the other hand, can you use the Famous Mackie Level Setting
Procedure to set the input trims on all inputs, with real music,
assign the channels to the main L/R bus, raise the faders to their
"unity" position, raise the masters to the unity position, and not
have clipping on the L/R bus?

I'll bet you can't.

[Scott Dorsey]

Michael R. Kesti wrote:
Scott Dorsey wrote:

Not on my console! I can slam everything all the way up and it's all
fine. But it's true that different consoles make different decisions
about noise vs. headroom.

Can you apply perfectly correlated +14 dBu, 1 kHz sinusoids to all of your
mixer's inputs, assign all of those inputs to the the mixer's L/R buss,
unmute them, raise all of the faders to their maximum gain settings, and
not experience clipping on the L/R buss?

I have actually done this experimentally, while running tones into all
channels at +8 dBu for tape deck alignment. Have not tried it at +14
but it would be interesting to try.

The mix buss is a 600-ohm summing network, so there is a whole lot of
loss in the summing process. There are big drive amps on each channel
strip to source that current.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Scott Dorsey]

Eeyore wrote:
Scott Dorsey wrote:

Eeyore wrote:
Scott Dorsey wrote:

you design an analogue console so all the channels going full tilt with the
faders up all the way aren't enough to clip the buss amplifier.

Well actually you don't !

That'll clip nicely.

Not on my console!

What console is that ?

It's a Studio Z console, built around 1975 or so. 600 ohm summing busses.

But it's true that different consoles make different decisions
about noise vs. headroom.

No question. Some that appear very quiet may have useless headroom on the busses.

That's the thing about headroom. Most of the time, it's useless. But
sometimes you're glad you have it.

However that factor is indeed taken into account but has to
be compromised against noise buildup.

True, and a lot of folks have gone to active summing busses which can bring
the noise floor down at the expense of headroom.

No reason why they should to be honest. Did you have anyone specific in mind ?

Pretty much all of the low end consoles out there do that. The Mackie
is one of the more common ones I have been frustrated by.

Digital excels in this respect as long as it has very long word lengths.

And there is no reason _not_ to have very long word lengths for internal
representations today.

Given the right DSP hardware, absolutely so.

DSP hardware is cheap. That's the nice thing about the new digital world.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce]

On 27 Apr 2007 10:32:19 -0400, (Scott Dorsey) wrote:

No question. Some that appear very quiet may have useless headroom on the busses.

That's the thing about headroom. Most of the time, it's useless. But
sometimes you're glad you have it.

But of course, when you need it, it ain't headroom any more.

d

--
Pearce Consulting
http://www.pearce.uk.com

[Scott Dorsey]

Don Pearce wrote:
On 27 Apr 2007 10:32:19 -0400, (Scott Dorsey) wrote:

No question. Some that appear very quiet may have useless headroom on the busses.

That's the thing about headroom. Most of the time, it's useless. But
sometimes you're glad you have it.

But of course, when you need it, it ain't headroom any more.

Absolutely true!
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."