I have a 3DR quad running 3.0.1 and am trying to figure out why it flips over and crashes well into a flight just about every other time I fly it. At first I thought it was due to vibration but I think I have that pretty much under control. My graphs of AccX, Y, and Z look good for the most part with X and Y +/- 1 and Z between -9 and -10. Just before it flips though the Z values start rising and go well above 0 as or just before it flips.
I'm at a loss as to what is making it flip and could use some advice regarding things to check. I had a helpful dialog with Gary McCray to address vibration issues but am not sure how to diagnose something like a loose connection or bad motor. What else would make it just flip and crash? The frame, battery, and motors are solid. I seems that the copter flips more often when it's hot out but that could also be because I fly more when it's warm. It seems to flip is stabilize, RTL and Alt hold.
I attached the most recent rlog, tlog, and log files. The crash occurs around 96% on the tlog. Any advice is appreciated. It's getting expensive keeping up with repairs.
You're situation sounds extremely similar to mine. I would get it up in the air, I could loiter, circle, stabilize. About 3-4 minutes in the flight it would go haywire and flip, or a motor would stop just enough to make it dive to try to recover.
My issue was that I was upgrading the firmware on the apm without deleting everything and resetting before I upload new firmware. Search for my posts if you need details on how to reset the apm.
make sure your frame configuration is correct. x or +. and make sure accel is calibrated.
Yes, your vibrations look very low - well done.
As regards your log, at the point of crash, it looks like Roll/RollIn and Pitch/PitchIn diverge significantly. I'd probably guess at a hardware issue, as stabilise seems pretty reliable these days. Could be voltage perhaps - but thats not being logged.
I'd check all your connections - from battery through ESCs to the motors. Are they all soldered securely? Give 'em a good tug to check, re-do them if you have any doubts.
So you are able to take off and hover, fly, etc, intially fine. At some point or time into the flight the 'event' occurs and you crash by flipping. Is this the sequence?
If so, you should be able to duplicate this by restricting the movement of the aircraft and run it up until the event. A test rig or stand would do this for you and allow you to quickly check motors for seizing bearings, overheating ESCs, failing solder connections, etc.
The last episode we had of this behavior was a bad solder joint on the motor leads at the exit of the wire of the motor, NOT a builder added bullet connector.
My bet is on a bad ESC or solder joint.