Note: just as the release was going out an issue was found re support for the all Camera Gimbals on the APM1s (it doesn't work) and with 3 axis gimbals on the APM2 (2-axis works ok but not 3). If you're one of these people, please wait for 2.7.3 which will be out shortly.
UPDATE: 2.7.3 is now in the Mission Planner
Functional Improvements over 2.7.1:
As per usual PIDs are optimised for the 3DR/Jdrones quad with 850 motors and 10" props. If you're using more powerful motors/props and are seeing bad flight behaviour in stabilize, start by turning down Rate Roll P in 25% steps.
There is still some question on whether the default Roll/Pitch Rate P (0.175) is high enough (it was 0.185 in 2.7.1) and also some people feel the Throttle Rate should be higher (currently 0.300 but some say 0.330 or even much higher would be better).
Please feel free to report issues you find in the discussion below and/or add them to the issues list.
Thanks and enjoy!
Hey again All, thanks again to those who make this possible and for the hard work.
Some of you might know we fly APM on most of our big SteadiDrone rigs, but also our smaller QU4Ds which are specifically made for high speed fpv etc, since loading 2.7.3 the tability is very very very strange, i think it's got to do with the new YAW, but not sure, to explain it,
when flying fast forward and letting go of the stick, the quad ramps up into the wind (oposite direction to flight) very badly, to a point where it just cuts the motors!! I would have thought it would level out with perhaps a bit of correction, but this is crazy, even when tuning down all I terms to 0, it does the same thing, the main problem is when banking tight turns, the yew almost want to flip the quad, so i test bu simply hovering and then yaw failry hard right then left etc, the quad wants just about flip over, specifically on a right hard yaw, the back left motor gives it so much gas its crazy, this is pronounced when flying into a wind when banking, it just wants to flip, and often just cuts the motors completely ????!!!
Any idea of what this can be? I understand the stab is very good for most users here with small props/motors and normal flight, hovering etc, but for bigger motors, 2814 T-Motors with 12" APC props, it's just no good.
Any ideas ?
I've gone once again through the whole process, starting with flashing APM2, configuring modes, X+ frame and calibrating radio. Looks like the last step was skipped last time and that causes all the troubles.
Anyway, this is fixed now.
Maybe some flag in Planner that will check if all steps are done after erasing eeprom which will cause some warning that step A, B are not finished and flight will not be possible?
I am seeing bit of same kind of instability issues with yaw. My problems were much more pronounced when using ESC:s with different firmware. Forcing the yaw with hand would basically cause the copter to flip. I think these problems might be related to the issues I described here: http://diydrones.ning.com/forum/topics/motor-redundancy-with-pid-tu...
Hmmm... this is weird. Jason backed out most of the aggressive yaw control that I had put in, so it should actually be flying more similarly to how it did in the past. ie: sloppy yaw. That's what I've experienced so far.
I haven't noticed what you're talking about, but I haven't done much fast flight lately.
I think the issue there is that we directly connect the 5v atmega 2560 serial lines to the 3.3v gps. If you push the atmega up to 6v then suddenly 3.3v is very close to half and the 2560 gets confused about whether it's high or low.
anyway, glad you figured it out!
I did a fast and straight flyback in stab mode with my quad (Dji 450 Frame 1,2KG with 3300Lipo3s&FPV, Roxxy 2827/26 (930KV), APC SF10/4.7 / APM2.0/ "normal" Skywalker 20A ESC), then i let go the sticks for a test. It did the described ramp up and brought the quad to stop, impressive. The motors didn't shut down. Looks like a good feature. On the next fpv flight with minimosd in althold mode i did a sharp left turn with yaw/aileron/pitch command at 42 km/h (thats fast for me:) - normally i cruise 20-30 km/h ) and it was super stable - no instability observed. I have a groundrecorder vid of that flight. When fiddeling in stab mode in the backyard i observed a slight, unproblematic yaw jump when doing fast yaw right/left changes with no instability/flipping tendencies. All Pids at default only rateP lowered - i think it is 0.130 now (don't know exactly). Higher rateP are also ok but will increase the downwash shaking on descent.
Yes, it's the exit_mission that is kicked off when it runs out of commands. So looks like it'll land if your rtl_approach_alt is zero otherwise it will go into loiter
Could you post of logs if possible with the MOTORS log type enabled? I'll bet it will get better if you reduce the Yaw Rate P term quite a bit and maybe roll and pitch rate P terms as well. The yaw Rate's default is .250 I think..on my hexa i reduce it to .150.
Rob, Hopefully the AP_Curve stuff that will correct for the non-linear pwm->thrust curves we're seeing from most motors+escs will get us back to something like that yaw control you put in.
hey again Randy, I did reduce it, as well as a few other terms, but I did notice one motor mount was slightly (very slightly tilted) I straightened it out and pow, issue gone, I was amazed that a very slight skew motor can cause such a violent reaction. weird..
Thanks for the conformation Randy, That explains why I was exiting mission into loiter, when some others were experiencing auto_land.
I actually did do some testing regarding linearization of the the motors (longer explanation here: http://diydrones.com/forum/topics/linearizing-motor-thrust) but as I took some measurements of my own at a bench test, I noticed that the thrust is fairly linear in relation to the command given to the ESC. My guess would be that the esc:s function by controlling the actual torque given to the propeller, so it might result on fairly linear thrust.
Now my test setup was not all that perfect since the propeller was rather close to the table so the ground effect plays a big role but here is the data I got: