^^^
skotdee just posted a similar angle while I was typing this:
The smooth knob is clearly digital gain, as I've said before - at its zero point you get digital silence.
Maybe, probably, but your conclusion is not necessarily correct. Just look at the R-09. '0' input level mutes the input, yet no one would argue that the R-09's input adjustment isn't controlling the analog input stage. The R-09 has a digital adjustment of analog input gain. So does the R-44, except there are two input range controls and it is uncertain how the gain stages work which those pots control. Both knobs adjust digital control signals, which can be programed to mute the input or set it at various gain ratios. Those digital control signals can be used to adjust either analog or digital gain changes and that's where the mystery lies.
I agree that both the wording of the manual and the simplified circuit diagram in the appendix seem to indicate that the 'smooth' knob on the R-44 is doing a digital gain adjustment after the ADC. Yet also I've been told by circuit savvy people poking around inside this machine (who seemed quite impressed with Edirol's engineering, FWIW) that the R-44 uses a PGA or programmable gain amplifier and that the simplified diagram might be somewhat misleading. I assume the R-09 uses a PGA design too. I'm not certain what all the implications of using a PGA are other than eliminating sending audio signal though the gain potentiometers by instead adjusting a control signal to the PGA, eliminating the problems of noisy pot wiper scratch, pot linearity and matching, and analog circuit routing issues. I'd think such a design would make for easy/clean/inexpensive push button or rotary user controls and simplifies the analog circuit by keeping it all in the chip and on the board. But I don't know enough about that stuff to really say.
The short and long of it is that I'm back to accepting that I have no idea what is
really going on inside this thing. I've reverted to a practical view informed by usage alone. In the real world I've run the R-44 with the smooth knob at both 12:00 and at full clockwise and have made good sounding, undistorted, noise-free recordings both ways. Yet I can also create the condition where I can cause the input stage to clip (channel number blinks) before the meters indicate full scale if the center knobs are not at 12:00. Because of that I've tended to set the center knobs at 12:00 and adjust the clicky knobs for sufficient gain. That makes the metering accurate at least which is vitally important, and I haven't noticed a significant difference in noise. Regardless, I try to keep my eye open for blinking. I was mildly shocked to find that the R-44 doesn't have an obvious flashing red clip light like the R-09 when I first received it.