I just posted a new experimental version of Arducopter with new PIDs and control laws for Yaw. It's download only for now and I wouldn't recommend trying it unless you can rigorously test it. Please be careful. I've flown this in Stable, Simple, Acro, Loiter, and Alt Hold with no known issues. But you never know! This version is not yet compatible with the mission planner.
New PIDs - I rewrote the control laws from scratch to add a PI Rate function. The end result should fly nearly identically to the current version. The nice detail is that we can use NG PID values for easy transition!
Before: -> After
Stabilize P –> Stabilize P (Use NG values, or 8.3 x the older AC2 value)
Stabilize I –> Stabilize I (Stays same value)
Stabilize D –> Rate P (Stays same value)
–> Rate I (new)
I added a new value – an I term for rate. The old stabilization routines did not use this term. Please refer to the config.h file to read more about the new PIDs.
I added the framework for using DCM corrected Accelerometer rates. Code is commented out for now.
Added set home at Arming.
Crosstrack is now a full PID loop, rather than just a P gain for more control.
Throttle now slews when switching out of Alt hold or Auto modes for less jarring transitions
Sonar and Baro PIDs are now combined into a throttle PID Yaw control is completely re-written.
I had a really good flight last night.
Did you disconnect ESCs before firmware uploading ?
It's a known problem and has been reported and discussed in the issue tracker :
No you must disconnect the ESC signal lines so that they don't receive perturbating signals.
see here in the wiki :