I just leave my setting at 256.
My system becomes unstable at the lower settings on busy projects.
And I stay with 44.1 because most of my work is headed for a CD so why fuss.
Because you can direct monitor at the interface there's no need to hear what comes out the other end of Sonar while recording. You only need to do this when using a guitar sim or if you want to hear a plug in effect live.
So I only use input echo on MIDI tracks, not audio.
I did use it once when a client wanted to hear reverb while tracking, managed to mix the direct and the delayed just so and he was happy.
The trick was crank the reverb send on the track but turn down the tracks level. This results in you hearing mostly the reverb. But you still need the direct monitoring mixed in so you can sing in snyc.
Track level has no bearing on the recording level so you can get away with that.
I now use a small mixer for monitoring so I can add reverb to the headphones.