Now I'm curious about the gain structure and performance of the line vs mic in on the R-09 as well. I've got to imagine that as a less complex Edirol machine it also just pads down the line in.
Back a year and a half ago when my R-09's line-in jack broke and people started reporting that as a weak point, I switched to running my external preamp into the still functioning mic input for about 6 months until I had a chance to send it back to Roland for repair.
To do so I just backed off the gain a bit on both the external preamp and R-09 to compensate for what I assumed was additional gain added by the R-09 mic preamp stage when using the mic input. I was pleasantly surprised to not hear any additional noise or degradation in signal quality. My assumption at the time was that line-in would be cleaner than mic-in on the R-09 and that I was essentially trading gain I had been making with the higher quality external preamp (going line-in) for gain made with the R-09's preamp (mic-in). I now realize that in all likelihood by goin mic-in, I had instead effectively eliminated the input 'pad' on the R-09's line-input, allowing me to back off on the necessary gain addition of the external preamp.
With the exception of one unusual (for me) and extremely loud show where I could have used a more padded input (I ran the external at 0 gain for the purpose of powering the mics only, and still ate up all the headroom in the signal chain with the R-09's mic input at a gain setting of 1), I had no problems running this way until I had the unit repaired. I at times thought it sounded 'cleaner' this way, but had no way of confirming this without going to the trouble of setting up a test. I was also still playing with finding the 'sweet spot' setting for my chain at the time so I was hesitant to attribute any perception of improvement to mic-in vs line-in without carefully eliminating the other variables. I only did casual critical listening, no noise floor analysis of 'effectively equivalent' gain settings that would yield identical levels of the same signal recorded though the line-in and then the mic-input. Now I'm interested again. Perhaps I'll go back to using the mic input more often.