I'm wondering if the max mic input is much higher than Tascam's stated -16dbV.
Upon further review this appears to be correct. I did some testing today (details below), to summarize briefly:
Running line in below 100 will result in chopped waveforms. Scratch this - turns out it was my source that was clipping.
Running mic in below 67 will result in significant distortion. But levels between 67 and 99 provide usable attenuation (or lack of gain?). Mic in appears to be able to handle a signal as hot as within 4-5 dB to what the line in can handle at 100. Taking your V3 results at face value, that would suggest a max input on the mic in of about 0 dbV.
Details:
I generated a 1 KHz sine wave with peaks at -0.1 dBFS in Sound Studio on my Mac. Went USB out > Lyrix Headstage DAC/headphone amp at max output > DR2D (recording 24/44.1).
First, for a reference level, I found the digital volume adjustment needed on my Mac's output needed to yield levels just shy of triggering the clip light on the DR2D (it's worth noting this happened with meters at ~ - 2 dBFS) at line in / 100. This required a volume adjustment of -6.5 dB.
http://homepage.mac.com/satterwill/Tascam/LineIn100.wav [warning - these sine waves are not fun to listen to, especially at high volume]
Then I ramped the digital volume control up to -0.5 dBV. This required line in be bumped down to 94 to avoid clipping on the meters. The resulting file sounds audibly different and has visibly flattened peaks and an altered Fourier analysis - http://homepage.mac.com/satterwill/Tascam/LineIn100.wavhttp://homepage.mac.com/satterwill/Tascam/LineIn94.wav While I suspect this level of distortion may not be blatantly obvious on more complex material, it does suggest there is no point running the line in below 100 and thus the lack of control over line in level while recording in dual mode is irrelevant - just set it to 100 and if you need lower levels, you need external attenuation.Upon further review, it looks like the Lyrix that was clipping, not the DR2D. Putting an attenuator between the Lyrix and the DR2D still results in a clipped waveform. By contrast, I was able to burn a CD of the testfile go DVD out > Naiant Tinybox (+4dB gain) > DR2D which clipped the DR2D at line in / 100 but recorded cleanly at line in / 97. So the line in may be able to handle an even hotter signal at levels below 100, but just how much hotter is unclear.
As to the mic input: First, I suspected based on the published max input specs that the max input of the mic in, IF AT 100, might be the published -16 dBV, i.e. 22 dB lower than line in's max. And indeed, putting the digital volume control at -28.5 yielded the target levels for mic in 100, with no distortion.
http://homepage.mac.com/satterwill/Tascam/MicIn100.wav But it turns out you can usefully attenuate the mic input with the internal level controls - but only down to 67 or so (I'll stick with 70 to be safe).
Results were:
-22.5 dB yields good levels at mic in / low / 88
-16.6 dB at 76
-13.8 at 70
-10.6 at 67
This appears to be the hottest signal it can take without distorting:
http://homepage.mac.com/satterwill/Tascam/MicIn67.wavThis is 4.1 dB lower than what gave me full scale at line in / 100. If that is +6dBV, it would imply a max input for the mic in about 5 dB lower, so +1dbV or 0dbV to be "safe".
Boosting the signal to -7.4 dB on the digital volume control to get "good" levels at mic in / 65 resulted in significantly more audible (and visible) distortion than the line in / 94 :
http://homepage.mac.com/satterwill/Tascam/MicIn65.wav Just for fun I tried - 4.2 dB (i.e. 2 dB hotter than the line in could take), at mic in 61 it got even worse:
http://homepage.mac.com/satterwill/Tascam/MicIn61.wavIt occurred to me that distortion and max input levels might be frequency-dependent, so I reran the -10.6 playback into mic in 67, with a sine wave at 42 Hz. It was clean:
http://homepage.mac.com/satterwill/Tascam/BassTest.wav