The 83dB reference level calibration is based on SMPTE RP 200 as recommended by Bob Katz. However, reference level calibrations vary based on room size. ATSC 85 provides the following recommendations based on SMPTE RP 200. Probably most here have a room under 5000 cubic feet (~142 cubic meters).
Media Center uses 83dB so unless you can change that target, that's what you have to calibrate to.
However, since 1% = 0.5dB in Media Center you could calibrate to say 73dB at 80% volume which, assuming everything is behaving linearly, should be equal to 83dB at 100%.
The current issue is if you wanted a reference-level signal to play at 63dB, which means peaks of 83dB as the tones are -20dB, you would either need a setting of 120% in Media Center, or a way to specify your own target. (either because your setup can't go louder, or you want to limit the volume from the PC via the receiver)
Thinking about it some more, allowing values above 100% would probably be the best solution rather than allowing the reference level to be adjusted. It should save a lot of headaches.
I wish Loudness wan't tied to Internal Volume Reference Level, but had its own setting. Music has no Reference Level. If I use Loudness for music I will typically lower my Reference Level by 10 dB or more to prevent excess boost in the bass and treble.
I don't know how you could do it any other way. The loudness contours are specific to the playback volume. (phon on these charts = dB)
Not deviation from the reference, but absolute to the level coming out of your speakers.
It is assumed that mastering is done at 83dB, so reference level = flat.
Our sensitivity to low frequencies is diminished at lower volumes, and follows these
equal-loudness contours.
These contours show how loud frequencies have to be so that they are perceived as being the same volume.
With an 80dB 1kHz signal, a 20Hz signal needs to be played back at about 120dB (+40dB) to be perceived as being the same volume.
When you reduce that to a 60dB 1kHz signal, 20Hz needs to be played back at 110dB. (+50dB)
If you assume that the content was mastered with 83dB as the reference level (let's use 80 for simplicity) then that when you reduce the playback volume to 60dB (-20dB) the bass has to be pushed up by 10dB to compensate.
This is not a "bass boost" it's what is required for the two tones to be perceived at the same volume.
If you reduce the volume even further to 40dB (-40dB) you need to boost the lower frequencies by 20dB for them to be perceived at the same volume.
I would assume (I haven't checked) that Media Center simplifies the loudness contours and boosts the lower frequencies by half of the reduction (-15dB volume = +7.5dB lfe) and does something similar for high frequencies.