Need advice on routing a singer through reverb while tracking

Describe your input chain - how is the mic signal getting into Sonar?

I'd cruise Craigslist for a second-hand reverb unit, like a vintage Picoverb or Nanoverb. They can be had for as little as $20-40, since those old 12-bit reverbs aren't exactly hot sellers anymore. But they sound good enough for the singer's headphone mix. 

I do it all in the box, with SONAR's mixer :-) So I can't comment on OTB solutions. And there are multiple good ways and multiple bad ways to approach it? I'm just thinking, if there's a way you can do it, using the tools you have? (yes I would render the tracks and use only the bounced instrument/other vocal track for her) then you're just working with the
 
1. The backing track
2. Her wet (which doesn't get recorded)
3. and Her dry (which does get recorded)
 
Then you can delete the backing track, the wet track never really existed, and now you have her dry, when she "felt it" wet, which I agree helps some singers considerably. If it works, couldn't hurt? If it doesn't, you can go the external route. 

smallstonefan
Wow, thanks for all the great advice guys! sorry for the delay in responding - I've been down 4 days with a cold - AND doing a complete re-org of my studio. Jay Tee I was really excited when you mentioned the sends and I just looked them up. Unfortunately, they get printed so that's a no go. Unless I can use the send without the return? I'll have to try that - that might be the easiest route.
 
Bit - during my re-org I realized I have a Boss SE-70 that I use solely for the shimmer effect for my guitar. I have now routed that in through my patch bay and I'm sure it's got a decent enough reverb patch to use strictly for monitoring.
 
The Band - I do have a TON of stuff on these tracks so to get decent latenency with a VST I suspect I'd have to bounce down as you said, which is an option. If I can route the Send out without needing a return (not breaking the signal path) that could be a good option. It doesn't seem right intuitively to me though. I also have a Yamaha mixer with a send on it. I'm now thinking I can:
1. Live monitor the singer's channel with zero latency using the 1200f. Route this to Out A
2. Send the full mix to Out C/D (a stereo pair).
3. Run all three signals to the Yamaha mixer.
4. Put the SE-70 on the Effects Send of the mixer.
5. Plug the singer's headphones into the mixer.
 
Logically, this works for me and I'll give it a shot. If you think this won't work or there are gotchas, please let me know...

That's a variant of my suggestion, it should work well, and yes, the key is bringing the wet signal back into it's own channel input, not the FX return.
 
I have 32 hardwired ins on the system here, but only 18 channel strips in the DSP mixer,  and 6 of those DSP strips are devoted to FX returns. One pair for a comp/EQ loop, in case I have to take the edge off a pumped up performance, and two pairs for 2 different hardware verb's, basically one for the "room" everybody plays in, the other for an individual effect like verb and/or chorus on a guitar part.
 
I specifically do not record the effected signals for the main take, to preserve max options down the road, and opnce I am sure I have the dry signal in the DAW, I go ahead and record the effected signal anyway, why not?
 
The templates I use in recording always have my main busses, Verb, Delay, Drums, Bass, Guitar, Keys, and Vox preset, with sends to Verb buss, which has a plug that closely mimics the hardware verbs, so there's no fuss, or embarrassment over "naked" Vox on playback right from the start.
 
Even so, I still have full flexibility on mixing, as there's always a fully dry track recorded to "start from scratch" with.
 
The DSP mixer on the Profire 2626 is especially nice in this regard, there are 7 stereo sends, in addition to the stereo mains. Send 1 and 2 go to hardware verb 1 and 2, Send 3 goes to EQ and Comp, and 4 thru 7 feed ins on a 4x4 headphone amp.
 
This gives me 4 discrete monitor mixes in the DSP realm, with fully adjustable FX, and which can then be blended or soloed for 4 additional monitor mixes, to a total of 16 sets of headphones with 8 different mixes.
 
The key to achieving this was in lucking out with the 7 sends on the 2626, adding the 4x4 headphone amp for $175, and extensive graphical documentation on my part to avoid hunting for signals during a busy session. Put another way, if you have several DSP sends, a small investment in a monitor mix/amp and good signal flowcharts will open up your monitor system exponentially, and in my opinion, a robust set of monitor options is perhaps The Key to capturing great performances.
 
I'll go out on a limb here, and place this on equal footing with room treatment. Few other areas of effort will improve results more than control over the instant feedback loop achieved by room acoustics, AND monitoring flexibility.

FWIW it is possible to add 'comfort' reverb/echo 'in the box' without suffering latency problems. The trick is to send the singer a direct monitor feed from the audio interface and not from the recording track.
 
Set up a pre-fade send from the recording track to a bus effect with the reverb/echo on it. Make sure the effect is 100% wet & 0% dry and mix it into the monitor feed to the singer.
 
Enable input monitoring on the recording track so that it will send the incoming signal to the reverb bus. Pull down the track fader so that the latent version of the singers vocal is not fed to the monitor.
 
This will record a dry vocal and will be able to subsequently add whatever effect the song needs.
 
The fact that the reverb effect is delayed slightly by the latency of the interface is not relevant - just think of it as a bit of free pre-delay :). Your singer will not notice it as long as your buffer settings are 'normal'. Their own voice is not at all delayed, which is the important thing, because you are using a direct hardware monitor feed
 
Of course, an outboard device is more convenient, but not essential.