Ted, you can automate the gain nob... by starting with the gain, you should get another 18 DB of precision in your fade.

The downside of using the gain is that as you fade it, you'll also alter the threshold going into any compressor on the track/bus or any other dynamics processing relying on a threshold or input signal for its processing, which can & will cause some rather unpredictable results, so I'd stay clear of that if I were you.

There is more than enough precision already built into the track fader/envelope combination, you just have to set it up properly

I have not noticed this as I usually do a final fade of the mixed WAV file using Audition to trim the begining and end.  It does a nice smooth fade to 0 with no cliff.

To be honest, I've never had a problem using Sonar's automation to get the exact type of fadeout I want at the end.

'S'-curves are my favourite, where you combine a fast curve with a slow curve

I wouldn't even consider using the Main Outs or Master Buss faders for this - I'd do it on a track by track basis.

I wouldn't even consider using the Main Outs or Master Buss faders for this - I'd do it on a track by track basis.

 
That's a rather long winded approach for a master fade out though?  Are you doing that because of the inherent innacuracies of a global master buss fade, or because you want to make relative level changes on the way out?

Had this problem too but as mentioned above the trick is a good "S" curve.  In fact you get the best results if you treat the fade very carefully.  I usually zoom in on the automation (On Buss or Master or whatever) and when I start to hear inaccuracies in "what I actually expect to hear" then I edit the automation using the draw tool.

Granted it can be a little tedious if your in a hurry but with this approach you can get the perfect fade.  No question. 

Wood67

I wouldn't even consider using the Main Outs or Master Buss faders for this - I'd do it on a track by track basis.

 
That's a rather long winded approach for a master fade out though?  Are you doing that because of the inherent innacuracies of a global master buss fade, or because you want to make relative level changes on the way out?

Definitely the latter Wood.

And besides, how many tracks are actually playing at the end of a song?

You might want any single instrument to fade out before others, so track by track is the way to go (for me)

This may sound a tad thick of me, but can someone explain the S curve, as I'm not sure i can visualise how that works.

Wizrd - You'd need to have at least 3 nodes for the fade.  The fiirst at the start of the fade, the last a -INF and then place a third somewhere in the middle. Start with a linear for each point to point.  Then modify the first so it is a 'fast', and the second so it is a 'slow' - or vice versa depending on what you want.  That will result in an inverse 'S' shape (twisted through 90 degrees anticlockwise I suppose!) for the fade curve.

I'm going to try some of these tricks over the weekend and see how they turn out.  Most of the time for me the main body of the fade is fine, it's just those last few db's where it jumps.

From what I've been experiencing...it doesn't matter what type of curve I use.  When it gets to the "next to the last step" before -inf.....I hear a definite dropoff instead of a smooth fade going from that step to -inf.  I think that is what Wood is referring to also.

I'm gonna try rendering out the file without the fade and doing the fade in Sound Forge and see if that makes a difference.  It's just easier if you can keep the whole project within one program.  That way if I want to do a small change of the mix later on, everything else (including the fade) will be exactly the same.

As I stated before...it's not critical....just something I noticed with this long fade.