microphone noise specs

Quoted noise specs are more than a little misleading.

Here's an explanation from David Josephson:

Many modern mics with such low noise specifications (don't use the term "noise figure," this has an entirely different meaning) achieve these numbers by allowing the capsule to have a rising frequency response in the mid band, and equalizing it to "flat" with a filter that happens to match the commonly used A-weighting curve. And, reducing the mid-band damping increases output and reduces the amount of noise contributed by the damping air. Two cost-reduction strategies for the price of one, let the funny sound of a midband scoop-out filter be forgotten in the hype of lower A-weighted noise spec numbers.

The main noise source in modern mics is still the damping resistance of the air around the diaphragm. Noise levels easily achieved with reasonably selected stock low noise FETs are on the order of 4 or 5 dB less than the noise from the capsule.

As an exercise for the reader, note the amount of very low frequency rumble and high frequency hiss that is evident in various mics you know. Now go look up the way most mic makers specify noise (rms, slow averaging, A-weighted) and see whether this even measures the rumble and hiss.

Kurt
Thank you for that!

I know they don't treat the entire frequency spectrum equally when they publish noise measurement results, but it never occurred to me that the rise in the frequency response would have such an effect, although it's blatantly obvious once it's pointed out.

John

Good callback their Kurt.

I put virtually 0 stock in published noise figures. I've had fantastic mics with noise figures well into the 20s and then the noisiest mic I've EVER used was an ADK with a spec of "14 dB of self noise." (Same as the Schoeps)

The fact is, even a "noisy" mic will not really be all that noisy when used in real world situations. Even when not, you're more likely to notice digital noise floor or physical noise floor before microphone self noise.

J.

I disagree with these statements. The ISO self noise spec is valid, and the A-Weighting still represents how annoying a noise spectrum is to us humans.

If a mic manufacturer tweaks his mic self noise floor to cause the spectrum to be the inverse of the A weighting curve then its still going to sound quieter to us and the listener when recording. The ear is less sensitive to rumble and hiss. I wish people who report this stuff would name names, regarding this tweaking of mic noise. Who does it, Rode?, Neumann? If it was such a crime and damaged the sound of these mics, we would hear it.

In the absence of a more detailed and thorough noise spec involving say true loudness measurement, something that most manufacturers and users will never understand let alone measure and report properly, the A-weighted ISO self noise figure is still one of the most useful mic specs to take particular note of.

In my experience and for my gear, the mic self noise is still what dominates the noise floor, it is much more significant than preamp noise or digital encoding noise floors.

DavidSpearritt wrote: The ISO self noise spec is valid, and the A-Weighting still represents how annoying a noise spectrum is to us humans.

OK. I find that my experience often differs from this.

If a mic manufacturer tweaks his mic self noise floor to cause the spectrum to be the inverse of the A weighting curve then its still going to sound quieter to us and the listener when recording. The ear is less sensitive to rumble and hiss.

Within limits, yes. But noise is only one component of a very complex set of characteristics. Once it has been reduced below a fairly obvious level the other components take on more significance.

If it was such a crime and damaged the sound of these mics, we would hear it.

Many of us do hear it, that's the point.

In the absence of a more detailed and thorough noise spec

Something that the quoted author has been championing for at least a decade.

the A-weighted ISO self noise figure is still one of the most useful mic specs to take particular note of.

Maybe.

OK. I find that my experience often differs from this.

Are you saying that you have found many examples where one spectrum that sounds louder and more annoying than another even though it has a lower A weighted magnitude?

Within limits, yes. But noise is only one component of a very complex set of characteristics. Once it has been reduced below a fairly obvious level the other components take on more significance.

Could you elaborate on which characteristics and what components, you refer to?

Many of us do hear it, that's the point.

Are you saying there is one or more mics that clearly sound inferior for music recording to you, primarily because their noise floors have been artifically tweaked? Which mics? I am most interested.

Something that the quoted author has been championing for at least a decade.

Granted, David has been most enthusiastic about getting the spec more detailed and properly defined. But I fear its a lost cause. I sure hope he succeeds. Certainly if anyone can do it, the AES in conjunction with ISO will do it.