[Adrian Tuddenham[_2_]]

I have just bought a Tascam DR-05 recorder and have been trying to
calibrate the mic sensitivity. As a first step, I calibrated the line
input against the machine's attenuator with a 1Kc/s sinewave:
http://www.poppyrecords.co.uk/other/images/Graph1.gif

With the exception of the last point at the high-sensitivity end of the
scale (which was probably upset by noise in my test set-up) there is a
very good correlation between the on-screen readings and the input gain
setting, giving a slope of 0.5 dB per attenuator step.

For this part of the test, I was able to use the overload indicator LED
as an accurate level detector.


I then set about correlating this with SPL levels, using filtered white
noise centred around 1 Kc/s in a room without any serious acoustic
problems. I used an SPL level meter almost touching the Tascam to set
the amplitude of the noise. For this test, because of the random nature
of the noise, I used the -12dBFS marker on the Tascam's level meter,
rather than the peak-indicating overload indicator.

I plotted the sound level points against input gain (for -12dBFS
indication) on a second graph:
http://www.poppyrecords.co.uk/other/images/Graph2.gif

The points do not appear to lie on a line with a slope of 0.5dB per
attenuator step; and a line corresponding to the overload point passes
significantly below 120dBA, which the manufacturers claim can be handled
without distortion.

I have checked that the SPL meter's attenuator is reasonably accurate by
using a calibrated attenuator in the white noise circuit. There are
obvious sources of error in a crude experiment like this but the
graphical points were reasonably repeatable to a greater degree of
accuracy than the deviation from the 0.5dB-per-step line.
It almost looks as though some sort of compression is being applied to
the mics or their amplifier. Does anyone have any 'inside knowledge' of
whether this is likely to be the case?



--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk