In order to finally get absolute signal levels I use a common function gene=
rator that is setup to provide a 1 kHz sine wave at a signal level of 1.09 =
Vpp, which is equal to -6 dBu (one could also use another field recorder or=
a computer soundcard if the exact output signal level is known - a common =
multimeter for instance would allow to measure that level).
I connect the microphone input to the output of the function generator thr=
ough a home-made 56.5 dB attenuator, which means that the signal level at t=
he microphone input is -62.5 dBu. I record this signal for a few seconds at=
the maximum gain setting of the recorder. I then disconnect this reference=
signal and terminate the microphone input using a 150 ohms resistor and re=
cord this silence (the inherent noise floor) as well at the maximum gain se=
tting.
In the analysis software (I use Avisoft-SASLab Pro) I first band-pass filt=
er the recorded .wav file from 50 to 18 kHz and calibrate it based on the r=
ecorded -62.5 dBu reference signal using the calibration tool of the softwa=
re. I then measure the rms of the noise signal, which provides the unweight=
ed EIN value. If the frequency spectrum of the noise floor is flat, I subtr=
act 2.1 dB from the unweighted EIN to get the A-weighted EIN.
One can also use other software that can calculate rms values (including t=
he free Avisoft-SASLab Lite software for instance, which does however not a=
llow direct calibration). You just had to subtract the rms [in dBFS units] =
of the reference signal from the rms [dBFS] of the noise level and add the =
absolute signal level [dBu] of the reference signal used.
Regards,
Raimund
|