Can anyone tell me what the ideal dither settings are when one is mixing down from 24/44.1kHz audio to a 24/44.1kHz mix, and using a final limiter plug-in (in this case, FabFilter Pro-L)?
 
Do I use the dither settings on both Sonar and the plug-in? Do I use dither on just one? If so, which one? Or, is no dither necessary since the final output is 24 bit - i.e., is dither necessary when going from 32-bit float mix state to a 24 bit file?
 
This is very confusing . . .

And which type of dither to use in Sonar?
 

If you're not changing bit depth there's no need to dither.

.

But - if the project is recorded in 24 bit, but uses 32-bit float during processing, efx, etc., is that - or is that not - a reduction in bits?

pdarg
But - if the project is recorded in 24 bit, but uses 32-bit float during processing, efx, etc., is that - or is that not - a reduction in bits?

Great question, I don’t know the answer but will speculate that since 32-bit float has a significand precision of 24 bits that it is unnecessary to dither unless you are exporting to 16 bit. 

pdarg
But - if the project is recorded in 24 bit, but uses 32-bit float during processing, efx, etc., is that - or is that not - a reduction in bits?

But you're not rendering in 32 bit -  as far as my limited knowledge goes, 32 bit float is only used for internal processing - not for actual rendering (exporting, bounce to track etc)
 
You only dither when you render down from 24 bit to 16 bit. 
 
Only ever dither once, right at the end when you're exporting (rendering) a final mix,

Rendering can also be done in 32 or 64 bits. (Edit > Preferences > File - Audio data, see "File Bit Depths"). Any bit reduction can introduce noise, and dithering is an attempt to "shape" the noise to make it less audible.
 
A different question is whether noise in the least-significant bit of 24 bits could ever be audible over a properly leveled signal. Therefore, would dithering down to 24 bits be distinguishable from no dithering?

Dithering from 24 to 16 bit doesn't make an obvious difference to not dithering much of the time.

As for choosing between different agorithms, try a few and see if they make any noticable difference to each other. I suspect the audible differences between algorithms won't exactly leap out at you.

In other words, don't sweat over it too much.

I only ever use dither when I go from 32bit FP or 24bit to 16 bit. I also only bounce out at either 24bit or 32bitFP, because for the most part the track is only going to be Mp3...meaning I keep the highest possible bit depth till the very end. My advice is, unless the track is going to be bounced down to 16 bit...no dither.
 
For what it is worth, dither setting 3 is for highly dynamic music...classical, some forms of rock and folk. Dither setting no 2 is for less dynamic music, rock, EDM that sort of thing. You shouldn't need the other settings theoretically. 
 
Ben