I'm wondering if any of the 990 users out there have the need for the test tones used in the 990 in an extended version, to be made available as .wav files.

I was able to acquire (record to a digital recorder) the test tone taken from the 990's outputs (not really a tone, but it's easier than saying "band-limited shaped random noise") and extend it to 120 seconds' length from it's original 17 seconds.

I did this by making a copy of the original, and then performing some math operations on the copy, and then joining it to the original, and then repeating those steps; lather, rinse, repeat. I'll describe how I did this later in the post (for the technically inclined).

If there is any interest from 990 users, I could post these to my FTP site in calibrated fasion (i.e. - 30 dBFS) so that when played on a CD player they will generate the same SPL as the test tones do. The longer file length just makes life a bit easier in the sense that if you want more averaging time for your sound level meter, you would have it. There is one big down-side, and that is I can only generate one or two-channel format files, so you couldn't use the files to calibrate all of your speakers.

Mind you, the only way in which the test tones can be guaranteed to play back at the same level (as those generated by the 990) is if the following conditions are met:

a) The CD player / media device upon which these files would be reproduced is connected to your 990 via its optical or S/PDIF output; analog inputs do not necessarily guarantee this.

b) If you were to download the test tones and burn them to a CD, you have to do so being careful NOT to invoke normalization or any DSP-type effects in your burning software, nor can you artifically alter the scale factor. Doing so would result in the test tones being at something other than - 30 dBFS, and thus, they would not be the same dBFS magnitude as the test tones proper (and thus, would not play back at the same level as the test tones).

The procedure for creating the extended version goes something (well, exactly) like this:

1) Acquire the test tone from the 990.

2) Edit the original 17-second file down to 15 seconds' length (not necessary, but I wanted the end result to be exact multiples of 15 seconds as that is a nice round number).

3) Make a copy of the 15-second file.

4) Reverse the sequence of the abscissa values, that is, turn the file around (for example, the instantaneous value observed at 7.2 seconds in the original file becomes the instantaneous value at 7.8 seconds in the copy; the instantaneous value observed at 4.0 s in the original becomes the instantaneous value at 11.0 s in the copy...etc).

5) Concatenate the copy and the original ('join' the copy to the end of the original file - this yields a file of 30 seconds' length).

6) Perform a half-cosine interpolation at the values immediately before and after 15.0 seconds of the 30 second-long file; this is necessary as there's no guarantee that the file will end on a zero-crossing. The interpolation ensures that no 'click' or 'pop' will result at the 'junction' of the two files, and the half-cosine method makes the interpolation much more gradual than would be achieved using linear interpolation.

7) Save the 'new' file, which is now 30 seconds in length, and free of any pops, clicks etc.

8) Repeat the process (60-second file results), then repeat it again (120-second file results).

For laughs, I also synthesized a version of the test tones by creating a white noise file of the same overall power as that of the test tones, and using those to create a filter that could be applied to white noise. Subjectively, the synthesized test tones sound identical to the actual ones, and when you look at their power spectra, they are basically line-on-line - and they have the same overall dBFS value.

Anyway, if someone has an interest in the files, I can make them available. If not, then I'll just use them as needed. If I have time, I'll update this post with some images showing the spectra of the test tones as well as the synthesized test tones.