Some thoughts and questions.
We often talk about 24 bit ADC but in reality it would
probably be pretty difficult to actually achieve full
24 bit resolution.
24 bit would give us a range of 144dB, as someone once
said that would be like the difference in the level of
a bird singing over a mile away and Concord taking off
right in front of you.
But are there not other considerations we should think
about?
-126dB is the inherent noise of a 600 ohm resistor. So
then the very best we could expect to achieve would be
21 bit resolution? But then it would take some
excellent circuit design to achieve that, would it
not? If we consider a real world device that say
supplies phantom power, there is probably a 'chopper'
circuit in there radiating a lot of electronic
rubbish. Then there may be a CPU as well, again a
device prone to radiating electronic shash. So would
it be more realistic if we thought of 24 bit system as
being able to resolve no more than 18 bits in a real
world scenario?
The nearer we get to those low level bits we are
attempting to measure some incredibly small voltage
changes. Which would require for some exceptionally
quiet circuitry, would it not? Is it possible to build
something which can supply such quiet circuitry
necessary to be able to resolve 18 bits let alone 21
bits?
If we then look at an example of say a microphone
pre-amp with say -126 dB noise. If we then apply a
modest gain of say 40dB we now have a noise figure of
-86dB, to help keep the maths simple lets say the self
noise of the mic is 6dB. That would give us a noise
figure of -80dB, would it not? A signal easily
accommodated, if we take care over how we mod, in a 16
bit systems 96dB.
Obviously if we used a 24 bit system that 'possibly'
resolves to 18 bits then we would have a bit more lee
way with our mod levels I would hope.
But 40dB of gain is 'modest' adding anymore gain and
we could easily accommodate it in a 16 bit system
could we not?
So is the moral of this tail when recording 16 bit to
mod as loud as you dare? Then in the editing and post
processing reduce the level to a more realistic
volume? But if you are able to record 24 bit you can
have a little more lee way in how loud you mod? But
don't expect to resolve much more than 18 bits?
Phil
--- John Hartog <> wrote:
> My earlier statement, "whenever you can, record in
> the highest bit
> depth your equipment is capable of" is too general.
> That would only
> make sense if the components of the recorder are
> fine enough to not
> induce noise to the point it negates benefits of the
> higher bit depth.
> Still if you are unsure, it is best to err on the
> side of highest
> potential quality.
>
> We can assume by the Sound Devices example, at least
> with their
> recorders 24 bit rather than 16 could make a
> significant difference.
> Still we haven't heard any field examples of this -
> or have we? And
> even though numbers from Raimund's recorder tests
> (thanks for putting
> that together Raimund) suggest some other recorders
> might be up to par
> =96 still the proof is in the field.
>
> So let's have some real world examples.
>
> John Hartog
>
>
>
___________________________________________________________
Yahoo! Mail is the world's favourite email. Don't settle for less, sign up =
for
your free account today http://uk.rd.yahoo.com/evt=3D44106/*http://uk.docs.=
yahoo.com/mail/winter07.html
|