So do you think we can conclude the loiter flyaways and bad RTL direction are due to compass auto calibration going crazy ?
Don't think so.
These may actually be a symptom of something deeper. However it definitely makes it much more likely....
There seem to be quite a few routines going 'wrong' at some point. GPS readings (particularly with some GPS's), the compass cal, the handling of the accelerometer in loiter, altitude hold during yaw, etc. etc..
I've been playing with the code with auto-cal turned off, running without accelerometer for loiter, and while the loiter is nowhere near as good, this hasn't (so far) glitched, even using a Mediatek GPS. Wanders quite a few feet, and is slow catching gusts, but doesn't change height when it corrects position.
for me its quite evident in the FPV recording that the compass is making the quad fly in crazy directions. and it makes perfect sense as a compass reading wrong heading by 180 degress would create a positive feedback loop, sending the quad accelerating in the wrong direction like a bat out of hell.
i think that so far its very good news that all the people with manual compass calibration have had no fly aways.
Really nice Emin... if you make them for sale please contact me in private! :-)
That is right, I mentioned 5 deg from my past Quads experience.
Can I do tilt on just one boom? Rear one.
ad me as friend so i can send you msg...i will explain what u need..
Ok, that wasn't clear you did an auto-land. It sounded like you were in Stab, and put the throttle all the way to the bottom, but it still wouldn't stop. That's practically impossible, which is why I assumed the Xbee might have been spoofing the ESC signal.
yes ACC should be calm as I have balanced motor cans and props and have 2 layers of moongel under APM case.
Torque does not turn it fast enough and I do not really want to increase YAW RATE_P to the sky. Idea is to be able to maintain it airborne with one failed motor and get to home location using RTL. Currently it barely does it. Depending on which motor failed you can yaw one way without loss of altitude and another way with spiral down. I will address this with motors tilt. That will give much more authority to yaw.
Good catch on Rate_D. It was tuned to the extreme in the still air. But in the wind the arm which is down the wind behind the airframe starts oscillating a bit. Turbulence I guess. I will tone Rate_D down a bit. Wind was strong that day... 25+ km/h I actually shoot the video from the ground using smartphone. That was a flight minutes before the one you saw from on-board cam.
Even after 12 flights, it is a good idea to try and maintain that fear which can give you an edge during flight. We must train ourselves like fighter pilots to know when things can go bad, and be ready to calmly react the right way as soon as things do go south. Having well over 50 successful flights in a row, I still keep a hair trigger on the stab switch whenever I loiter near anything. Even with manual compass, a solar flare can still take out your GPS... happened to me after over 20 good flights in a row, and could've hurt somebody bad if it decided to turn towards the flight line 10' away. Since that I realized no quad can be trusted 100%... not enough redundancy given all the possible failure points... more reason to keep that fighter pilot attitude.
Graham Dyer, default rate_D at 0.010 felt so soft for my frame/motors/props... I may try 0.050 or 0.045.
I can't see any difference tuning rate_I with the knob on CH6. What is the proper procedure for rate_I tuning? Can you please point me to a page if you know which one is right.
Very nice Emin. Those copper elbows look more robust than the usual plastic pieces. Do you machine your own parts? These brushless gimbals make me regret not buying a small mill. ;)
i mesured coper part and they are lighter than 8(6 i dont trust) tube block with carbon and screws....this is prototype i made for myself from old gimbal and parts i found around my house..we in croatia are not so lucky to go in store and by peaces u need,i have to order almost anything or pay 5(yes 500% more) times price here...i was thinking making gimbals but u simply canot compete with China..they are good and cheap..here is example for all of you interested in bigger gimbals
or 99$ and 199$