I have an Apogee Ensemble that runs with Logic and a long time ago I set it up so people were monitoring themselves directly from the Apogee Ensemble's mixer Maestro. The advantage of this is that because they're monitoring off a hardware device as opposed to software there is virtually no latency. With this set up I mute the channel I'm recording on in Logic or the performer would hear the latency free signal from the hardware AND the delayed signal from Logic's latency. This sounds like a slap back echo and is very distracting!
This set up works great and allows me to have my buffer size up quite high while still giving performers low latency while recording. A problem arises though when someone wants reverb in the cans. Years ago I figured out a little trick. If you click on the "sends" rectangle that says what bus your send is going to some options will come up. Here's a picture:
By default a send is "post fader" meaning the volume of the fader effects the volume going to the bus you're sending the signal to. If you select "pre fader" the signal going to the bus won't be effected at all by the channel's volume fader. What you do next is turn the volume of that fader all the way down so you don't get any of logic's latency signal in your headphones. You're still going to have some latency with the reverb or delay but it's never an issue. You can just consider it some extra pre-delay. :)
When the send is set up to be "pre fader" you'll notice it turns green as opposed to blue:
I should mention that when I do this I always set up a separate "record track" and slide the vocal onto another track for playback. There's too many variables to adjust if you don't do it that way.
Catch me on the sites: