I Use a MAIN OUT BUS and a MASTER MIX bus :
The MAIN OUT BUS is the FIRST bus: easy to get to to adjust.
The MAIN OUT bus controls the monitoring volume sent to the audio device, via sonar's main outs.
The MAIN OUT bus does NOT have "effects." And no envelopes.
The MAIN OUT bus has monitoring tools: Voxengo Span, vintage meter, Panipulator (required),...
The second bus is the MASTER MIX bus.
All other Buses output to the MASTER MIX bus.
Guitars to Guitars All, and then to Instruments All then to MASTER MIX
All Drum/Purc buses to Drums All, and then to Instruments All then to MASTER MIX
Vocals Main, and background, to Vocals ALL then to MASTER MIX
The MASTER MIX bus is output to the MAIN OUT BUS.
The MASTER MIX bus is used for bounces. All final effects are in the bin.
The MASTER MIX bus may have envelopes for fades and misc. adjustments.
I add at least two blank tracks labeled MIX1 and MIX2, to projects.
These output to the MAIN OUT bus.
NOT directly to the sound device.
Not to the MASTER MIX bus.
I solo the mix track to listen to the mix(s). When listening to the MIX tracks there is no need to bypass the effects/eq, or change the volume setting, of the MASTER MIX bus when listening to a mix track.
And because the MAIN OUT BUS volume is the same, I can compare, audibly and visually, the mix with original tracks.
Per CJ's insight, if the mix does not sound correct, before I do anything else, I use control-z (undo) to remove the mix clip, and the wav file from the disk. This means I have fewer orphaned audio files.
ALSO: I change the name of the mix tracks (add the date) before bouncing-to-tracks so the wav file will have a reasonable name.
I never look at, or change, Sonar's Main Audio Outs.