From observation the basic gain settings have much to do with how well, or not, GPS hold works. With high basic gains and the lower ATTI gains the large frames need, GPS seems to easily go into TB mode. Being able to lower the basic gains helps a lot, but two things I've found that seem to make the most difference, get the IMU measurements as close as possible, and get rid of as much RFI as possible. That would include things like twisting the motor wires for the length of the run between motor and ESC and eliminating as much non shielded and straight run wires as possible. Also, and I know some people doing builds right now won't want to hear this but, get the power distribution board out of the center of the frame and out from underneath the IMU. The IMU gives off enough RFI as-is, it doesn't need any help in creating more and having the main power distribution directly below in the center of things seems to do exactly that. On both the CS6 and my Droidworx AD6 I've gone to the same power disti setup and elimated a lot of excess wiring, both have seen corresponding improvements in GPS performance. Coincidence? Maybe, but I'd be more willing to believe in a reduction of interference so the GPS can do what its supposed to.
Ken
