Hi all,
I'm interested in all your experiments and tests about microphones.
I'm always searching for a very sensitive but quiet recording rig. I
did some comparisons among the mics I have and these are the results:
I did the test at night, in my house, quite silent because of old
walls that are 60cm thick! External noise come in only through
windows and doors, that are modern. At night you can barely hear
trains passing at about 2km and some airplanes.
I used an electronic metronome that makes a quiet bip.
Microphones at 2.5 meters of distance, on a thick carpet to avoid
multipaths on the floor.
Recorder Sound Devices 722, set at max gain with all microphones,
recording to CF to avoid disk noise.
With the Rode NT1A the bips peak at -10dB and my breath is recorded
several meters away!
The EM23s I got from Klas are powered through a box with a 9V battery
inside. The other microphones were powered by the 722.
The EM23s appear to be 4.5dB more sensitive than the Rodes.
The AKG appear to be 2.5dB more sensitive than the Rodes.
Then I analyzed the recording in the spectrum window of Adobe
Audition 2.0 with a large FFT size (65536) and by averaging across
few seconds of recording.
After compensating for the different sensitivities, at 12kHz I get a
noise level of -112 with the Rodes, -101dB with the EM23 and -97dB
with the AKG. But a significative difference is that the noise floor
is flat for the AKG and Rodes, while in the EM23 is not flat: -96dB
at 6 kHz and -110 at 40 kHz.
Then I tested the CoreSound Mic2496 with the Rodes; the results are
pretty good, but they are not significative because I own an early
unit without the box. I just have the internal electronics. Thus I
can't be sure my measure reflect the performance of current units.
In the same recording conditions I got bips at 12dB less than with
the 722 because of the lower gain available.
After compensating for the different gain, the noise floor is -109dB.
Not bad if compared with the -112dB of the 722.
In the spectra the bips peak around -39dB at 4 kHz. This means that
the real SNR I got at 12 kHz is 73dB the best case, 58dB in the
worst. The bip level was around 65dB(A) at 1m, measured with a cheap
chinese phonometer. Of course measures made at 12 kHz on a spectrum
can't be compared with the measure made with the phonometer.
If you like I could assemble a mono file with all the samples, before
and after normalization.
Gianni
--------------------------------------------------------------
Gianni Pavan
Centro Interdisciplinare di Bioacustica e Ricerche Ambientali
Universita' degli Studi di Pavia
Via Taramelli 24, 27100 PAVIA, ITALIA
Tel +39-0382-987874
Fax +39-02-700-32921
Web http://www.unipv.it/cibra
Email
|