--- In Walter Knapp <> wrote:
> Whether or not you like the term, gain is often what you are going to
> find in published specs. One has to understand all the things that go
> into that terminology to work with it, but that's not hard.
>
> If you want to really confuse folks start talking entirely in voltages.
>
> And then get into impedance.
It isn't a matter of not liking the term, it is a matter of using it
in the appropriate situation. Gain is used for specifying mic preamps
when you know the target output level. But it isn't used when testing
a portable recording device, because what is relevant is the level of
the recording and not the output level.
For example, lets say I put a -60dBu signal from a mic into a recorder
and set the record level for -20dBFS. Now, lets say this recorder has
both -10dBu (often -10dBv, but I don't want to confuse the
calculations) RCA output jacks, and +4dBu XLR output jacks, and it is
calibrated for 20dB headroom so that -20dBFS =3D reference output. So,
I've got 50dB gain out of the RCA jacks, and 64dB gain out of the XLR
jacks. Same tone (or bird), same recording level, different gain numbers.
The voltage gain needed to drive the internal A/D is the true
electrical gain, but we aren't measuring that, and this doesn't matter
to the end user. The signal-to-noise ratio from either output will be
similar, since the mic and preamp noise from this small an input level
will dominate over the noise floor of the electronics. Of course, I'll
use the gain at one of the outputs to make the noise calculations, but
the point is that gain will vary depending on the output level of the
device.
So, in my example, one recorder with +4dBu outputs has 64dB gain.
Another has 60dB gain, because they've referenced -16dBFS instead of
-20dBFS to reference output. Consumer devices with unbalanced outputs
are all over the map. Same bird, same recordings 20dB below full
scale. Now, I dump all these recordings onto my computer. Now the gain
and the output level are meaningless, because it depends on what my
computer's audio interface puts out.
What I'm really trying to find out from my measurements is: When I
record, for example, a hawk at 200' with a Sennheiser model x mic, how
many dB down is the mic noise and how many dB down is the
preamp/electronics noise when recording with the Zoom H4 vs the
Micro-track vs the Sound Devices vs the Sonosax recorders. And then
express it in pictures and graphs, instead of just silly numbers. Stay
tuned!
> > The acoustic test level of 74dB or 94dB SPL at 1 meter is a fixed
> > reference level for testing microphone output and frequency response,
> > as virtually all microphones can handle this, and it swamps out normal
> > background noise so you don't need to be in a silence chamber. As
> > microphone gain and response normally don't vary with acoustic level,
> > these test levels should be valid for any situation, from birds
> > chirping to loud music. These are sound levels are for measuring
> > microphone output and response, and don't relate to the electrical
> > levels used for microphone preamp noise testing.
>
> I've found the assumptions above to be less than accurate if one gets
> into recording ambiance. There you are not so much recording the calls,
> but the silence in between the calls. You are a very, very long ways
> from the standard test conditions.
I'm not sure what you are trying to measure that you feel lower levels
would be more accurate. These test levels are used primarily to
measure output level and polar response. Polar response is a result of
the mechanical design of the microphone, and I don't know of any
scientific data that shows that the physical properties of the air or
the microphone would vary with acoustic level (except maybe at
extremely high levels, where you might get acoustic overload or air
turbulence). The output level is measured at a standardized reference
level of 94dB SPL so that mics can be compared. It should scale down
linearly to low levels -- if it doesn't, then the mic would have to
have significant distortion.
What could make sense might be to measure microphone distortion at
varying levels, from very low to high. The problem though would be to
find a transducer (speaker) that had much lower levels of distortion
than the microphone, even at very low acoustic levels, and speakers
are notorious for having high levels of distortion.
> I'd think you could not accurately test a mic's self noise without a
> soundproof chamber. Certainly not the very low self noise mics
preferred for nature recording.
Of course. Except, that for some mics, like the Schoeps, you can
replace the capsule with a test head containing a capacitor to
substitute for the diaphragm, so you can measure the self noise of the
electronics. I don't know if the noise is any higher with an actual
capsule, but you'd need a soundproof chamber to determine that.
- Adam
|