Hi David,
I'm now a bit confused. You say that if in ideal quiet conditions we
can observe a noise level rising according to the rule 6dB/oct then we can
assume we are recording the thermal noise and thus our recording system is
good.
However, many recorders have converters (e.g. sigma-delta converters) and
noise shaping algorithms that produce an increase of noise with frequency.
Probably we need first to check if the noise produced by the converter is
flat or not. How to do this ? By shortening the input ? or by lowering the
gain to zero ?
Gianni
2014-02-18 15:46 GMT+01:00 <>:
>
>
> > What exactly is the thermal noise level of the air ( I would suppose it
> would vary with temperature) and how is it measured and defined?
>
> Greg,
>
> It is the noise generated by vibrating air molecules at normal
> temperatures.
> It is about the zero phons level, but what makes that difficult to define
> is
> that the level rises with frequency at 6dB per octave. This is why
> weighting
> is used. Weighting curves go through zero at 1kHz, and mic hiss gets
> annoying at higher frequencies which is a problem when defining noise.
>
> A problem with A-weighting is that this is based on a 40phon human hearin=
g
> response, not zero phons. A more accurate noise weighing is the ITU-R 468
> noise weighting which is better matched to the ear at noise levels but it
> peaks at around +12dB which gives "poorer" figures for manufacturer's
> sales
> blurb.
>
> Quality mic manufacturers often give much "poorer" figures than those
> given
> for cheap mics because they are measuring something different. I've seen
> excess noise figures based on the mic capsule being disconnected. They
> sometimes get away with this by calling it "self noise".
>
> The lowest possible excess noise mic figure is 3dBs caused by air
> vibrating
> on the rear of the mic diaphragm as well. The record mic excess noise
> claimed by Bruel & Kjaer is 5dB, but for practical mic figures look at
> manufacturers like Sennheiser.
>
> Thermal noise varies with temperature above 0Kelvin or -273C, but in
> practice this variation is small at normal temperatures. Temperature also
> varies preamp thermal noise so this mostly cancels out.
>
> > Then how can the air 'noise level' be specified relative to the standar=
d
> for used for microphones?
>
> It can be calculated theoretically, but unless a manufacturer defines the
> base level it uses and the weighting curve used, they can make this up.
> Sometimes 0phons is used as a base level with no justification. You can't
> make comparisons using different methods of defining the noise measuremen=
t
> method.
>
> > Is it equivalent (nominally) to a mic with a self-noise level of 1 dB o=
r
> -3.2 dB or what?
>
> There is no standard specification for "self noise". Some mics can get
> hissy
> when hot. Disconnecting a capsule is meaningless as you want to know the
> noise level in its useable state. If you connect a passive capacitor in
> place of the mic capsule, that also has its own thermal noise, but if you
> leave it open circuit, nothing is comparable. Excess noise over the
> thermal
> noise level of the air and over the whole recording system, and weighted,
> is
> what counts.
>
> If a spec is too good to be true, ask why. :-)
>
> With a very quiet background or under a pile of bedclothes, you can
> sometimes see, on a spectrogram, a rising level in the higher kilohertz.
> If
> you are lucky and with an excellent mic, this is the fundamental thermal
> air
> noise rising at 6dB per octave. If you can see this in your recordings,
> you
> are doing well.
>
> David Brinicombe
>
>
>
--
Centro Interdisciplinare di Bioacustica e Ricerche Ambientali
Universit=E0 degli Studi di Pavia
Via Taramelli 24, 27100 Pavia, Italy
http://www.unipv.it/cibra
http://mammiferimarini.unipv.it
|