More testing, 1/8" input
Line In, 0 gain: -80.43 dB -90.81 dB
Line In, Half gain: -74.03 db -84.14 db
Mic in "M", 0 gain: -72.28 dB -82.43 dB
Mic in "M", half gain: --66.84 dB -76.47 dB
In each case, 10 db more noise in the left channel.... hmmm...
Yeah . . . I just got a Microtrack. I did some voice and classical guitar recordings, about 2 feet from the T-mic. M setting, level all the way up, 27dB boost off, 24/44.1, and man, the noise is pretty well unbearable. Noise is -52dB, average signal RMS for the guitar recording is -25dB (-6dB peak).
I thought I would be able to clearly hear the clock ticking in the room, but it's completely lost in the noise.
So I did a bunch of tests. I put 1.2K metal film resistors across the 1/8" inputs. Best case noise (L, min level) was -79.8dB, worst case was -55.5dB (H, max level). So the T-mic is noiser than a 1.2K resistor.
For the TRS I just shorted the inputs and for the left channel I got -99.1dB at L + min level, -74dB at H + max level. The right channel was 3dB to 10dB noisier (opposite of the figures quoted above).
Clearly the -100dB noise figure quoted by m-audio is best case.
Clearly I don't have to bother making 24-bit recordings. Looks like I'll need to post-process with noise reduction software. Looks like I won't be keeping the Microtrack after my current project.
I wanted an R-09 but it wasn't shipping yet, and probably still isn't. I hope it's cleaner than the Microtrack.