I'm on an older MacPro3,1. I have two audio interfaces, as well. I use a AJA Kona LHi for my video io, and I have a MOTU Traveler firewire for my system audio. It has phantom power, so that's what my mic is attached to. So, in that way, our systems are similar.
For editing inside Premiere I have Adobe Desktop Audio set to the Kona card. Map Output For: Desktop Audio. Playback Preferences > Audio Device: Adobe Desktop Audio (and the Kona as my video device). These are the settings that give me the most reliable audio performance.
SO... for the VO recording, I have to change the Adobe Desktop Audio to the MOTU Traveler in order to get the mic input. And as I mentioned, I'm not recording along to picture; I'm just recording wild VOs. I could see how using a separate interface for the audio and the video could create some issues for recording audio to picture.