So many questions. If the vox recording is low, how low is it? What are you using for the other tracks. Any other acoustic recordings should be low, too, unless you have a very soft voice. If you are using loops or something like Z3TA (which always goes into the red, it seems), a normal acoustic recording will sound limp incomparison, unless you turn them down.
It sounds more like a mixing problem than technical. If you are recording a strong signal (almost clipping as you describe) but you can't hear the vocal you need to turn down everything else for the mix. If you are listening on headphones while recording and have the "thru" button button on in SONAR it passes through the signal recorded to the output. If you also have your interface passes the signal to the monitoring chain you will hear whatever you are recording twice and it will be louder, realitive to teh backing tracks, while recording.
A preamp will help with a low-output mic or impedence mismatch. Esp. buit-in preamps will usually only have 50 dB of gain. Combine that with a weak mic and a soft voice and you have to beg the input level, and most interface preamps will get noisy and crap out if you belt one word. Also, if you are eating the mic to get level small headmovements will affect the sound quality, including proximity effect.
But you say that you are recording a strong signal - it is only weak on playback. My suggestion is to look at the mixing part. lower the faders on all the backing tracks all the way down. Put your vocal at unity (0 on most boards, virtual or real). Then bring the backing faders up, starting with the rhythm tracks. Keep them low so they don't mask the vocal at all, then bring up the other instruments. Most of the time the other tracks will be at -12 dB or less.
@