Posted by: "jpbeale"
> Thank you for the link and the comments, which are well taken. Yes,
> certainly I would need to calibrate the mic separately to determine
> sensitivity, and that is more difficult, especially to obtain accuracy
> over a wide frequency range.
It's certainly going to need well calibrated measuring equipment to even
get started. Hardly seems worthwhile, manufacturers figures are good
enough. The next "test" from there is to get out and record your normal
subjects and find out just how well it does. The test numbers simply
won't tell that. Far too much enters into actual recording.
If you want to go down the route of producing accurate and unbiased test
numbers on mics it's going to cost to get everything in high enough
quality and precise enough calibration. Cost so much to do it right it
should be considered a separate hobby as it will only be mildly
connected to actual field recording experience.
> The manufacturer data often lists self-noise as a single number in
> dB(A). I think ITU-R 468 would be better
> http://en.wikipedia.org/wiki/ITU-R_468_noise_weighting
Sennheiser and some others do provide this measure. The mic data website
has a entry for this where it's available.
Your big problem is that folks are used to the A weighted measure. Start
sticking in CCIR data here and there and you just will create confusion.
I've found that the A weighted measure seems closer to experience in
nature recording than the CCIR method. Generally the better a mic is A
weighted, the better it will be for nature recording, at least in this
aspect of mic characteristics. I've not found the CCIR to agree near as
well. I used to follow both measures, but have pretty much settled on
paying most attention to the A weighted numbers. To put it another way,
CCIR has probably as big a flaws as A weighted.
> but what I am really interested in is the actual sound of the noise
> and its frequency spectrum. I am also not convinced that all
> microphones of a given model have exactly the same amount or character
> of noise.
I've made this point here repeatedly, with little interest shown. The
character of self noise is very important in nature recording. There
simply are not microphones quiet enough for the quietest sites. Even
with just fairly quiet sites the mic's self noise is audible in the
recordings. If it's of a character that is not very noticeable or
intrusive that's far better than, say a mic that sputters, crackles or
whatever.
If folks want to put up tests of lots of mics for comparisons they
should provide recordings that make the self noise by itself clearly
audible for comparison. It does not take elaborate lab equipment to do
this. And it's a critical issue. This would be far more useful than
questionable comparisons of test numbers which seem to be what everybody
seems to be doing.
BTW, one of the reasons why I and others prefer MKH mics is the smooth
and unobtrusive character of their self noise. Not the only reason I
might add.
You are right that a given model of mic may vary in it's sound between
individual mics, not just the self noise, but even how it reproduces the
sounds you want. One of the advantages of the high end expensive mics is
that they are far more consistent in this than budget mics. Thats part
of why they cost so much. The differences between individual MKH mics
for instance are so tiny that they will be unnoticeable in actual
recordings of nature. When they are noticeable it's time to send them in
for Sennheiser to check out.
Note in evaluating the sound of a mic, self noise or wanted signal you
really want to read references from the Mic data website library.
Particularly read the impedance paper to find out just how hard it is to
get unbiased comparison samples. Just one aspect of how hard it is.
And no matter how many tests or comparison samples you go through, when
you get out in the field making actual nature recordings it will take
the same amount of time to learn to use each mic to it's best. I take
the approach that once I have a mic in hand it's time to get out and
start that learning process. Don't waste time testing when you could be
learning.
Walt
|