naturerecordists
[Top] [All Lists]

Re: testing mic self-noise

Subject: Re: testing mic self-noise
From: Walter Knapp <>
Date: Tue, 27 May 2003 16:03:49 -0400
cmursic wrote:
> Pardon what may be an incredibly basic question, but how would I go
> about checking a microphone's self-noise level?  Is there some way to
> do it by using, perhaps, an audio editing program like Sound Forge?
> Thanks in advance
> Caesar

To do it accurately takes a soundproof room and calibrated sound sources
and metering. What's usually done is to expose the mic to a 94dB sound
source, set the output meter to read 0dB, cut the sound source, get
another reading, which is the self noise. Anyway, that's the simple
explanation. It will be reported as the actual dB reading, or may be
reported as the difference in dB. Two standard sound profiles are used,
one is even for all sound frequencies (usually C weighted), the other is
a profile that matches human hearing sensitivity (the A weighted scale).
Typically the dBA spec will be lower than the other, and is usually
what's talked about.

Note that self noise is reported relative to a fixed sound level and
gain setting. In actual use you will rarely have your gain set such that
94dB gives exactly a 0dB meter reading. As you change the gain on your
recorder, the actual self noise level will vary in how it corresponds to
a environmental sound level. What you really want to understand is this
correspondence. ie the mic as you have it set up at a particular time
gives you a noise floor that corresponds to a noise of ?dB in the
environment.

For instance if you have the gain set so 84 dB is 0dB on the meter, then
the self noise would correspond to a environmental sound level 10dB
above the mic's rated self noise level, or at least that approximation,
the variation may not be even but probably works fairly well for small
differences from 94dB.

This setting of 0dB for 84dB sound would not be a uncommon one in my
work with frogs. By sound meter measurements, many of my closer frog
sites are in the 74dB range or so for their peaks. Since I set for those
peaks to be at -10dB on the meter, that's like setting 84dB for 0dB on
the meter. So, for instance, my SASS/MKH-20, which has a self noise of
10dBA would record at those settings with a self noise floor that
corresponded to a environmental noise level of 20dBA. Meaning I'll have
about 54dBA of dynamic range available in this example. Now think of
using a mic that has a self noise of 25dBA, like the Sony ECM-MS957. The
self noise would now correspond to 35dBA environmental, leaving only
39dBA dynamic range for my recording. Now think about some sites where
the frogs are quieter, and my sound meter measures peaks at 55dBA. Or
ambient backgrounds that can read 30-40dBA on my meter.

You can see why it's less important to those recording a rock concert
at, say 125dBA peak sound. Your gain will be set way down, the self
noise will then be down too. Lots of dynamic range even with a noisy
mic. (assuming your mic can take that sound level)

Please note these are quick and dirty examples, we have not even taken
into account differing mic sensitivity, but assumed it the same. The
sort of thing you can use for rough guidelines and practical nature
recording, but not probably technically exact. And to know the numbers
you would need a calibrated sound meter in the field anyway. I have such
a sound meter, and do keep track of the sound level, but don't get into
these sorts of self noise numbers. You can listen to tell. Simply put,
the more you raise the gain, the higher the mic's self noise will be in
terms of sound in the environment. Technically it's self noise did not
actually change, you simply amplified the signal from it, which has the
same effect when you talk about corresponding to sound in the environment.

A quick and dirty method, which won't give you calibrated results but
still is useful is to simply take the mic into a quiet room, bury it in
pillows or such like to remove any remaining sound and listen to what
you still get off the mic. Since you will probably be feeding it through
your recorder to do this, you are really getting the combined self noise
of mic and recorder, but that may be more useful anyway. You can get a
fairly good idea of the recorder's self noise by disconnecting the mic,
though it's pre may react different with no mic. And the self noise of
the pre may vary with gain setting in ways not directly connected to the
gain. (Translation, the pre probably has a "sweet spot" where it will
give the best results, typically at mid to 2/3 gain or so)

Note if you have several mics for which the self noise is known, the
pillow method can give a fairly accurate estimate of a unknown mic. If
you account for sensitivity differences.

It does pay to check out the self noise of each mic you have, learn what
it sounds like. It's quite variable. A desirable sound for self noise
would be a smooth hiss, but some mics sputter or crackle. With a smooth
hiss, the self noise can actually be audible and not upset the listener.

Walt




________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU