My quad has been flying pretty well with 3.1RC7 but i have a persistent issue with yaw and roll pitch instability that is most obvious when flying in auto or guided mode.
The quad flies great in stabilize mode, alt hold and loiter without major issue. The quad will accept and navigate to waypoints in guided mode and has successfully run auto missions. However, if the mission calls for subsequent waypoints with more than about 90dg difference in heading, the rapid yaw commanded by autopilot is immediately followed by a roll/pitch instability and a dramatic loss in altitude .
This has happened 3+ times, with both 3.0.1 and 3.1RC7. Usually i note the signature "wiggle and dip" in time to throw it back into stabilize mode and save it. Today i wasn't so lucky and the quad hit the dirt after i gave a guided mode waypoint at ~ 180 heading from the quad's previous commanded position. The quad went unstable and dropped 30+ft to the field below.
After replacing broken frame parts and checking over the system, i took it for another test flight in stabilize mode only. I noticed during this flight that dramatic yaw input from the sticks shows similar behaviour in stabilize mode.
Is this a tuning issue? Is my Yaw PID set too aggressively (I'm at default)? Is there something in my logs i should be looking for to determine what the root cause of this is? I've reviewed my tlog from the guided mode flight that resulted in crash and i can't see anything obvious.
my setup:
F450 clone frame, APM 2.5, xbee telemetry, 3700mah 3s LiPo, Turnigy 1000kV motors (D2830/11), GemFan 1045 props, canon sx 230 downward facing cam as payload. AUW is ~ 1400g. hovers well at ~45-50% throttle.
My tlog is attached. I can see the event, but nothing indicating what led up to it. You'll note that i accidentally left my telemetry on for a bit after the funny business...
Thanks in advance for any help!
:-)
Replies
Hi Guys,
Any update on the issue above ?
i do have the same problem.
Regards
Hi Daniel,
Did you advanced with this issue?
Daniel Sisson said:
Hi,
ArduCopter 3.1.2 on a 500 class quadcopter.
I also have a major problem with yaw during auto mode, the yaw is so strong that overall stability is lost on the Roll/Pitch axis, causing "fluttering" sometimes until it hits the ground.
I do not understand under which conditions requested yaw should be so strong that it overpowers the stability of the craft on other axes. IMO, Under no circumstances (requested high yaw rate or no) should yaw take precedence over roll/pitch stability of the craft... and yet.. it clearly does. It seems that roll/pitch stability should feed back to requested yaw rate and reduce yaw input if either are perturbed beyond a certain threshold.
I will attempt to reduce both RATE_YAW_P an more-so AHRS_YAW_P to reduce yaw affecting roll/pitch stability.
Video of instability in action..
did you find out what caused this ?
Hello did you solve that problem?, I having a drift issue when yawing, thanks.
Continued to investigate this and found the following:
decreasing Rate Yaw P solved the dip during yaw. Quad responds a touch more slowly, but no longer loses stability during hard yaw.
however, I'm still having an overall stability issue. I'll start a seperate thread for my persistent problems.
Brandon, did you post a seperate thread for that? Link?
I think I see a similar sort of issue in loiter mode with mine. Even in a nice steady hover in loiter, if I yaw the copter around it tends to drift from it's original position pretty dramatically. I dunno if this is just the compass getting confused and not knowing which motors to bring it back in time, however it does sound related. Maybe just try yawing in loiter mode and see if it drifts out of position too?
This does sound like your yaw rate is set too high. See: http://copter.ardupilot.com/wiki/tuning/
STABILIZE_YAW_P produces the input into the rate controller, so both affect the behavior. in this case, since you're not continuously oscillating, it sounds like STABILIZE_YAW_P is the best thing to tweak. If you reduce that, the commanded yaw rate will be lower, but you'll get a descent tracking anyway.
If RATE_YAW is too low, then you probably see a slight yaw oscillation taking place or the inability to achieve the correct heading. Both can also be verified in stabilize to see how the vehicle responds to commanded yaw inputs.