Whenever I've read about feeding multi-band quads I've only seen the two methods below:

1. Match to 50 ohm Z for each band with a 1/4 (3/4) wl coax stub, bringing all the stubs to a remote switch from which exits a single coax line to the station.

2. Tie all the driven elements together with ladder line or whatever from which you feed with ladder/open line back to the rig using a tuner/baluns in some combo to match to the rig. Sort of a fan dipole method except that you don't expect each element to present 50 ohms +/- at its resonance. (I do this now, btw, except that I use a balun and fixed network at the tower and then coax).

I've used both of these methods successfully, but I'm wondering if one couldn't do away with the remote switch in method 1, feed each band separately and use coax all the way? Thusly:

3. Match each band with coax stubs, which theoretically means each element presents 50 ohms +/- at the end of its stub. Tie all these to a junction from which exits your coax to the rig. Would it not work like a fan dipole, since for each band resonance there would be only one stub that is close to 50 ohms and that element would take the power (or at least the bulk of it)?