Please remember that only 16-bit stereo wav files can be written to a standard audio CD (although you can save mp3s and other formats as data on a CD).  That is one reason to dither when you export from Sonar.
 
I am sure others have their own methods, but when I am finally finished with a song, I export it three times:
 
1-As a 24-bit wav (no dithering) to save as the master, to send to the mastering engineer, and as a download for those who want high-quality (but large) files.
2-As a 16-bit wav (dithered) to write on a CD for listening purposes in different environments.  If a mastering engineer is making the final CD version for me, then I usually compress this one just as a trial version for my own use.
3-An mp3 to send by email or other methods for which a wav is too large.
 
We had a lively discussion on this forum recently about whether the particular dithering option in Sonar makes a difference a human can hear.  We all agreed that dithering in general can make an audible difference (though not always obvious).  When it came to the discussion about whether it mattered much whether you use Rectangle, Triangle, Powr-1, Powr-2, or Powr-3, I said no one could tell the difference.  Some people agreed and some did not.  I normally use Powr-2 but can't really justify that choice.

i did not read this article so I can't comment on it's validity but Figure 10: Overlapping Spectrograph Plots of Dithered vs. Truncated Sine Waves on page 9 shows a plot of a truncated sine wave with dither applied and no dither applied. compare that with Figure 3: Spectrograph Plot of a 24 Bit 100Hz Sine Wave shown on page 3
 
 
http://www.users.qwest.ne...ng/DitherExplained.pdf

brundlefly

tlw
If you're working with 24 bit audio then turn dithering off until you do the final actual conversion to 16 bit.

This isn't necessary. SONAR won't apply dithering to any bounce/rendering process unless there's a reduction in bit-depth.

I did not know this. I always set dithering to "None" when exporting to the same bit depth. Even with this knowledge I'll probably keep doing that because I'm an anal weirdo... unless of course there is a reason to NOT use that method.

Dithering is a like a placebo and a plaster. I also make 24 bit flac files for websites like soundcloud whomn support it I believe.

Well... Every day is a school day 
 
I've always left my dithering option as Powr-1 only based on my total lack of knowledge.
 
I work in 44.1/24 bit, export as a 24 bit WAV and then open in SF10 for normalising/compression. From SF10 I then save-as a 320kbps MP3 file for Soundcloud.
 
I do not burn to CD nowadays...
 
If Sonar doesn't dither if exporting as same bit depth then I guess I can just carry on as I am, but I'm tempted to do the same as Beep from now on......

The way I understood dithering, is that when you have a 24-bit file, and you have audio that resides in the lowest significant bits, and you downsample to 16-bit, you basically cut of the least 8 bits of the audio.
So, this means, that your "sound" in that LSB is lost (not that it would be much though). So, all audio in the LSB's are set to 0. When you dither, it adds a noise to your audio. Which might shift the LSB into the higher bits, and as such becomes "audible". In essence, it is the same as when you would have a number with digits behind the decimal point. Like 23.4 or 45.7. When you truncate, you get 23 and 45. Let's say you add 0.5 (the noise). The first case would amount to 23.9, and the second to 46.2. When truncating this, you get 23 and 46... So, the 'upper' part of the digits (the .7 in the second example) actually counted in your truncation.
That's how I understood dithering... Can you hear it, yes, but it would be very very subtle... Sometimes dithering is added to CD's as well (chesky records does this on some of their cd's) so that errors in the DAC's are minimized.