I'm working on an instrumental track which has a piano figure playing throughout. The other tracks in the project are all routed to an "Instruments" bus. The piano track is not. It and the Instruments bus feed into the Master bus.
 
I've set up a sidechain using the PC4K-S Type bus compressor on the Instruments bus, and it's being controlled by the signal from the piano track.
 
So far so good, but my idea was that when the piano part gets softer, the other instruments would get a bit lower in volume, and conversely when the piano is louder, the instruments would not be compressed. Overall, the piano track would never get buried by the other instruments, nor would it be too loud.
 
But after setting it all up, it seems that I've done the opposite. When the piano is louder, it compresses the instruments, and when the piano is softer, there is little or no compression of the other instruments.
 
Is there  a way to do this properly? Sort of an inverse sidechain?
 
Thanks.
 
PS: I've already applied compression to the piano track. But I was hoping to get a more dynamic, live feel by using a sidechain.

You want downward expansion, not compression, like the PC4K Expander/Gate. Set the bus level so that's the desired volume when the piano is loud, and the expansion ratio so that the soft parts are the desired level with respect to the piano when it's soft.
 
The Sonitus compressor can also do compression ratios below 1:1, which gives expansion. Although it's not really intended to do expansion, you may be able to wrestle the ratio, theshold, and output level to produce the desired result.

And once again the resident Maestro sends me off in the right direction.
 
Many thanks.
 
 

I'm a fan of this technique, too. However, I've never been happy with the results of rule-based volume tracking (e.g. this bus goes down 3 dB so that bus goes down 3 dB too) because it doesn't work equally well for all parts of a song.
 
There may be places where you want the piano's level to be much higher than everything else and other places where you want them to be more closely matched. Changing the ratio as the song progresses can add dynamic movement to the arrangement, such as gradually raising the backing instruments' levels. In a sidechain scenario that means another automation lane for the send.
 
You also have to be careful with abrupt volume changes that can cause unwanted pumping. If the piano is the featured instrument, it needs to preserve its own micro-dynamics independently. Leaving such decisions to a dumb plugin can result in a robotic-sounding mix.
 
The solution is hand-drawn automation. Tedious and time consuming, but it'll allow you to choose between sudden or gradual volume changes, whichever is appropriate for a phrase. It'll allow you to adjust the balance for best blend, e.g. a greater ratio on the verse and a lower ratio on the chorus. Or a gradually-reduced ratio as the song progresses.
 
So while it's tempting to experiment with exotic solutions, don't forget the old-school ways. They're still around for a reason.

Yes, it all comes down to using one's ears, doesn't it. And thank you for your input.
 
But it is a bit funny to think of hand-drawn automation as "old school." I started recording in 1980 on a TEAC A3440S (4 track Reel-to-Reel, which was recommended by Craig Anderton in his then cutting edge book, "Home Recording for Musicians"), and hand-drawn automation was at that time strictly science fiction.
 
The state of the art has come a long way.