Harvey Gerst in his excellent threads says that large diaphragm condenser mics are quieter than small, but how do they compare with dynamic? Is a dynamic mic quieter (less pre amp hiss) than a LDC? I'm having problems comparing mic specs because some mics have a db figure (e.g. -65dB, relative to what?), and some give sensitivity as mV/Pa. What is Pa? How do I compare a mic which says -65dB with one which says 1.5mV/Pa? And some mics say 15dB self noise - what does that mean? I need to know which mic is giving the least hiss relative to the signal from a given sound level picked up by the mic. Can anyone enlighten me?
Noise comes from many sources:
thermal noise, which is a physical property of anything with resistance, a few different types of what I'll call "active noise", which is not a real term, I just made it up for simplicity. Finally, microphones have another type of noise from random molecules bouncing off of their diaphragm, which is lumped in with their active noise in deriving a mic's self-noise spec.
Dynamic mics (most of them, anyway) have no active components (transistors, tubes), so they only have a small amount of thermal noise, which is small enough that we can ignore it. This is why you never see a self-noise spec on a dynamic mic.
For a dynamic mic, the limiting factor is always the preamp, due to their low sensitivity. On sensitivity, "Pa" is a Pascal, one of which equals 94dBSPL. Some mics are stated in dBV, some are stated in mV, but both measurements should be 1 Pa:
10mV/Pa = -40dBV @ 94dBSPL
20 * log (10mV/1V) = -40dBV
Personally I think that manufacturers that spec only mV and force customers to do the conversion should be flogged (they don't spec max SPL or self-noise in mV, do they? Huh? Huh? Huh?), but I am not in charge of international standards setting, so . . .
Back to dynamic mics, they have low sensitivity, let's use -50dBV/Pa which is actually high for a dynamic. You won't find too many preamps much quieter than -130dBV-A (A weighted, we'll call that "dBA"), so you can derive a signal-to-noise ratio of 80dB, which is equivalent to a self-noise rating of 14dBA (94dB - 80dB). That's about best-case scenario with a dynamic; an SM57 into a -125dBA preamp will be 10dB worse (24dBA).
So most LDCs will be quieter than just about any dynamic mic-preamp combo; many if not most SDCs will be too.
Electrets can be as sensitive as externally biased condensers; sensitivity is generally a design decision given that there is an amplifier inside every condenser mic (although LDCs will normally have higher output than SDCs). Since electrets are often smaller diaphragm than externally biased condensers they might be viewed as noisier, but that's not a rule, there are good quiet electrets (KSM32, KSM141, AT3035). The noisier electret mics will have a similar noise performance to a typical dynamic mic-preamp combo.
When you want to know which mic gives you the lowest possible hiss, select the mic with the lowest self-noise rating. For most condenser microphones preamp noise is not a major factor. For that to be true, the mic's noise floor needs to be well above the preamp's noise floor. Reversing the above calculations, let's say you have a -40dBV/Pa microphone with 8dBA self-noise. Its absolute noise is:
-40dBV/Pa - (94 - 8) = -126dBA
And let's say you have the preamp with the -130dBA rating. Those two noise sources will sum as follows (this is a bit complex*):
20 * log (((10 ^ (
-126 / 20)) ^ 2 + (10 ^ (-130 / 20)) ^ 2) ^ .5) =
-124.5dBA
So your preamp has degraded your noise performance by 1.5dB (-126 vs. -124.5).
What if you choose a more sensitive mic (let's say -35dBV/Pa) with the same self-noise rating?
20 * log (((10 ^ (
-121 / 20)) ^ 2 + (10 ^ (-130 / 20)) ^ 2) ^ .5) =
-120.5dBA
Now you've only lost 0.5dB compared with the mic's noise floor, which is a 1dB improvement over the less sensitive mic.
What about a noisier mic, say 14dBA (still -35dBV/Pa sensitivity):
20 * log (((10 ^ (
-115 / 20)) ^ 2 + (10 ^ (-130 / 20)) ^ 2) ^ .5) =
-114.9dBA
Only a 0.1dB loss, pretty much negligible.
From that we can prove our rule-of-thumb which is you can ignore a noise source if it is about 10dB quieter than your other noise source. If the noise sources are the same, total noise will increase by 3dB.
In conclusion, you will experience the lowest total noise by using a reasonably sensitive mic (>-40dBV/Pa) with a low self-noise rating (<12dBA is good for most people) into a preamp with a reasonably low "equivalent input noise" (the figure I was using above) specification, which should be -125dBA or less.
* and note that I have ignored the effect of noise spectrum by treating all A-weighted noise as the same. There are long and mostly uninteresting arguments about whether or not A-weighting in appropriate, but this post is long enough as it is . . .
