The Edirol R44 threads go back a long way, and it's not that easy to search for precisely what one is looking for. So in the interests of nailing this discussion of the function of the inner knob once and for all, for the current generation of readers, here's some links which I think will be persuasive.
First -
http://taperssection.com/index.php/topic,107018.msg1621904.html#msg1621904This shows that when the inner knob is all the way down, you get digital silence. That in itself seems to me to be persuasive - you'd expect at least some bits to be non-zero if this was an analog domain control. Also, as you increase the setting of this control, signal and noise increase virtually identically, which is what you'd expect from digital amplification (but not analog amplification).
This post also shows that as the "clicky" outer gain control is raised, signal to noise ratio improves (as you would expect of any mic preamp), and that when you increase from the position marked -14 to that marked -20, you get more than 12dB of gain but with 5dB less noise. This indicates that using higher gain with the outer knob provides the best signal to noise ratio.
Posts following that one did not dispute my findings.
Second -
Back in May 2008 I had this email from Roland Australia in response to some questions I'd sent them, which they forwarded to their R&D people in Japan. As usual with an email, where their reply quotes my questions, my quoted questions are indicated by ">".
"Hi Peter,
Got this response from Japan R&D:-
Dear David-san,
The block diagram describes only basic concept so that the customer does
not get confused.
> First, the manual says that the limiter operates in the analogue domain,
> but the block diagram shows it operating in the digital domain. In use,
> it seems to overcome preamp overloading, which would mean that it must
> be an analogue limiter - which is good - but clarification would be
> useful.
The R-44's limiter is analog and digital mixed like the R-4Pro. When you
turn on the limiter, analog gain goes automatically -12dB down for
prevention of digital distortion and after that passes through the
digital limiter. After that the signal level goes +12dB back up
again for matching total signal level.
*The limiter threshold is -10 dB relative to digital full scale.
> Second, the block diagram shows the continuously variable level controls
> (the inner knob, not the stepped sensitivity control) operating in the
> digital domain. The manual text doesn't seem to say whether that's true
> or whether the level is actually analogue domain. Again, it would be
> useful to be sure of how it works.
SENS knob ( outer knob ) is analog domain. ( stepped )
LEVEL knob ( inter knob ) is digital domain. ( continuous )
<About suggestion>
The R&D guy said that they appreciate for sending the suggestion.
It is impossible to be mounted the R-44 due to hardware and software
limitation.
But it is helpful to develop next generation product.
Please say our regards to the customer! Hope that all helps.
Kind regards,
David Lackey "
(The "<About suggestion>" concerned some suggestions I made for improving the product, not relevant to this post. Note also, outside the current topic, that using the limiter is pointless if the recording is going through a post-production stage later - just record at a lower level and use a finely-tuned limiter in your DAW to obtain the best, undo-able, result.)
So - we have very convincing tests undertaken (more than once actually) by myself, coupled with categorical statements made by Roland R&D Japan, vs statements
attributed to Oade which, when you think them through at any level do not make sense (eg the assertion that lowering the first stage of a two-stage amplification system and raising the level of the following stage is likely to result in an overall improvement of signal to noise), and which certainly do not correspond with the opinions of the vast majority of other users based on extensive experience, nor with the known behaviour of this device.
I reiterate that Oade are being reported as making incorrect statements about this device (causing end user confusion, not to mention wasting my time in refuting them) and this does nothing to enhance their reputation with me (for what that's worth) - as for others, well, up to you to draw your own conclusions.
If Oade are being misquoted, or quoted out of context, I'm sure this thread would be a good place to set the matter straight.