I suspect that the NCVT works by using the user as a reference, which under normal circumstances should act like a ground of sorts.
I think if you were isolated and touching a hot chassis with one hand and used an NCVT with the other it wouldn't light up.
If I'm correct, then putting an NCVT into any device would require a known good ground for it to be useful, otherwise it isn't any better than a neon bulb.
I could be wrong..
Jason, you are correct on all counts. A NCVT capacitively couples through your body to the earth, listening for the 60 Hz (or 50 Hz) hum. If you're standing inside a vehicle with its chassis elevated to 120-volts above earth potential (what the RV crowd calls a "hot-skin"), then a NCVT will not beep if you get near any of the metal within your Faraday cage. However, you can reach outside and point it at the ground and it will beep.
The NCVT manufacturers also warn that if you're standing on a tall fiberglass ladder you might not couple to the earth sufficiently and get a false negative. Some goes for testing inside a 3-phase power panel since it's possible (but I think unlikely) to find a null spot in the middle of the 3-phases that will give a false negative.
But building a NCVT into a tester is pretty simple as long as you intend it to test while someone has his hand on the side of the tester to provide the earth ground capacitive reference. I also have a short video of me with a Amprobe NCVT taped on the side of my Sprinter van with a little light-pull chain handing down to a few inches above the earth. Whenever the chassis/skin of the van is electrified with AC voltage, the NCVT will start alarming without anyone touching it. That's because the body of the tester is coupled to the side of the van, while the sensing tip to coupled to the earth via a paper clip and a little chain.
More to experiment along these lines later, but it sure is interesting.