franklieve has it spot on.

I use a compressor to smooth out vocals and also make them louder.

More exactly I'm trying to increase the gain of the microphone circuit so that the weak parts of my vocal range (i.e. bottom end / falsetto) get louder without the strong area (upper-mid) overloading the channel / amps / speakers. The purists may (with some justification) suggest that a lot of these problems can be reduced with good microphone and vocal technique but as a sit down pianist I've not got a lot of either to work with!

I set the compressor up to give about 10db of gain to the signal (use the noise gate to keep the channel quiet when no signal to prevent feedback). Then set the compression ratio to be about 2:1 (so the loud stuff isnt so loud any more). Set the limiter to cut in only to limit the "if I dont sing this note b?lls out I wont get it" overload situations. This also may allow a little more gain. Use "Soft Knee" limiting - "Hard knee" can be harsh when it takes effect (the knee is the shape of the limiting gain profile).

The "attack /decay" times are subjective. Too fast an attack means you dont get such a dynamic sound (the attack of the voice gets lost; letting the attack through before the compression takes effect is known as 'punch-through'). Too fast decay means the sound can audibly "pump" up and down. I think Ive got my compressor set to "auto" for attack and decay!

I've got my compressor set such that you can't really tell it is there - until you take it away and think "where's the vocal gone?"!

I can also concur that the "dbx" info is very useful.