Good stuff. I think we all run into the same design issues eventually. I would like to hear more about the TH-2. I will post some pictures when I can. My design is not based on any existing design. My goal was to build an entirely new and simple to fly VTOL based on off-the-shelf components with a fairly low end radio. In other words, something that would hover autonomously and that a kid could fly. While it was more complex than I expected, I think I've achieved phase 1, so I'm working on phase 2 which is to integrate some on-board controls to handle stability, transition to forward flight and motor failures. I can do some channel mixing on the Spektrum radio, but the VTOL really needs to hover without the radio.
My first prototype was a 16oz plywood airframe using 4 articulating ducted fans, but I couldn't get even half the published 8 oz thrust out of them, so it wouldn't get off the ground. My second prototype used the same GWS EDF-75 fans in a foam delta wing (F27 Stryker) which was much lighter but the fans still couldn't lift it.
My third prototype uses 4 SK400XT (cheap and plentiful at HobbyLobby) and 10x4.5 CR props which crank out over 20oz of thrust each!. I was using 4 eFlite S75 servos rotating 4 tilt shafts / wings, but it only took a few hard landings to strip out the servos and bend the shafts.
So currently I'm using 2 ball bearing mounted shafts for strength, 1 JR retract servo for tilting both shafts, and 4 Hitec HS65HB (carbonite) servos for elevons on each wing. I'd ultimately like to go back to 4 shafts and vector the thrust instead of using aerodynamic controls, but that will require even beefier shafts and more ball bearings (i.e. more weight) so that will have to be down the road.
Currently, I have 3 gyros; 2 roll and 1 pitch. The pitch gyro is inserted in the two rear motor circuits and I have 1 roll gyro on each of the front motors. This is the most stable configuration I've come up with yet, but these inexpensive gyros tend to creep pretty quickly. Rather than throw mor gyros at it, I'd like to make it more intelligent, so I was thinking about the Parallax Javalin chip controlling the PWM to the motors and servos based on x/y/z axis data from the gyros or an accelerometer. Anyway, I'm committed to making this work, but I hope to learn from the mistakes of others, since because it's too expensive to make them all myself.