I just ran a 'quick and dirty' test and came up with -29.7 dBFS. I plan on repeating this later tonight in a more controlled fashion, but the value seems reasonable to me. Once I finish, I'll post the final answer that I get up to my FTP site. The test tones are actually fairly band-limited random noise (I think from around 100 Hz to 2 kHz, but again, I'll go back and check).

I'm curious to read what the actual specified level of the tones is (in dBFS).

[Update: OK, so I repeated it a second time, and I got the same answer. As to how I came up with the value, here is what I did:

1) Construct a wav file that is 70.7% of full scale, that is, - 3 dBFS.
2) Connect a digital recorder to the front balanced outputs on the 990.
3) Select an arbitrary volume level on the 990. I chose - 20 dB
4) Play the known - 3 dBFS wav file from my Audiotron taking its optical output into the 990.
5) Set the gain on the digital recorder such that when playing the known - 3 dBFS file, no overload (or under-range) occurs on the digital recorder.
6) Acquire (record) 30 seconds of the - 3 dBFS file.
7) Invoke the menu on the 990 and go into the channel calibrate menu. Select Manual.
8) Acquire (record) the left front channel output.
9) Acquire (record) the right front channel output.
10) Take the SDRAM card from the digital recorder to my PC and edit the files for length to 15 seconds' length.
11) Open the wav file of the known - 3dBFS file in Spectra Plus (FFT Analysis software).
12) Select dBV (decbels relative to volts) and calibrate using the known - 3 dBFS file.
13) Post Process to compute overall RMS power. Result: - 3 dB
14) Post-process the wav file uisng flat-top windowing and 16384 lines in the FFT with 95% overlap.
15) Post Process to compute the overall RMS power of the test tone file using the same calibration factor as was used for the - 3 dBFS file. Result: - 29.97 dB

Note that Spectra Plus does not allow units of FS (full scale), however, the math behind dBFS and dBV is identical - the only difference is that for dBV the reference is 1.0. However, both follow the general expression used for dB, namely dB = 20 * log (v/vref).

If you want to see plots detailing these results, I have posted to .png files on my FTP site in a folder called test_tones_magnitude. To log on to the FTP site (from a browser), just type (or copy and paste the string below into a browser):

FTP://owners_990.immersifi.com:combfilter1.immersifi.com (Sorry!... I mistyped that... the correct link to copy into your browser is as follows:

FTP://owners_990.immersifi.com:combfilter1@immersifi.com - my bad).

If you are using an FTP client, then the logon credentials are:

User: owners_990.immersifi.com
Password: combfilter1
Host: immersifi.com

There are a bunch of other directories there, but the plots are in the test_tones_magnitude folder.
Anyway, it does indeed seem to be the case that the tones on the 990 (the total power in them) is essentially - 30 dBFS.]