Loiter Takeoff crash


I had a weird crash upon a succesful loiter flight and RTL landing. With in the same ARM I tried to take off at Loiter and the copter leaned right just after the copter took off and destroyed the main blades and tail boom.

I am kind of worried how can this happen when the copter make a good flight and next take off it crashes.

Version is 3.5.7. Here is the log file.

param file: trex700-357_crash.param (13.7 KB)


You experienced what is called dynamic rollover with a helicopter.

It appears to me that the #2 servo is reversed, #1 normal, the aircraft rolled right, it tried to correct for it, but it was already past the recovery point.

Whether dynamic rollover is caused by wind, taking off from a slanted surface, hooking a skid, whatever, only a human pilot has the skills to prevent it. Taking off in Loiter flight mode you have no control over the helicopter and the autopilot will not correct for it in time.

Complicating the issue with RC in UAV use is that many are running their headspeed way too low, which compromises cyclic authority.

I know many will say they take off in Loiter all the time with a helicopter. Personally I don’t do it as the autopilot is not even close to the precision I have flying the helicopter manually in critical flight stages like takeoff and landing.

Chris thanks for the valuable input first of all.

Heli was on flat terrain and was nothing to hold on to the skid for it to lean right. Yet I guess what you said was the issue where I didn’t have cyclic maneuver capability il Loiter. Duly noted for future flights.

No more Loiter take off.

Thanks again and I am ready to make another flight today.


Yeah, experienced helicopter pilots start flying them during runup (in full-size) before you pull pitch. It’s impossible to do that in Loiter with RC. In theory they should lift off clean, but sometimes they don’t. Even if it don’t have a hooked skid once it starts to roll over and goes past the recovery point there’s no bringing it back. Pilots that try to teach themselves how to fly even a ultralight helicopter with no formal flight training will do it every time:

Well it turned out a mechanical failure as servo 2 which is the 60 degree one failed as we fixed the heli and powered up. the swash didn’t get in position right as that servo stays at minimum position.

I am glad that didn’t happen when it was in air we only lost the blades and some screws.

But regardless Loiter I will not try at all for sure :slight_smile:

Definitely. If you would’ve been in Stabilize and it started to roll you would’ve slammed in aileron and dropped collective. Can’t do that in Loiter because it don’t respond on the ground and you only have collective rate control - don’t have real collective response. And collective is the only way out of a dynamic rollover.

Even when everything works right, as I said, I have way more precision control flying manually during critical flight stages than the autopilot can do. My takeoff procedure is to go light on the skids running at rated power and make sure everything checks, then lift off about 6" to a foot and hover for a bit to make sure everything is stable and checks out before going into flight.

Helicopters are more advanced and complicated aircraft than multirotor drones, and I’ve always said that trying to apply flight procedures used with multi’s is an accident waiting to happen with helicopters. They are not a pushbutton appliance like a “drone”.

I reviewed your log and it appears that the GPS was reporting that the aircraft was no longer at the same position, even though in reality the aircraft didn’t move. this is due to the poor accuracy of civilian GPS (15 meters). So once you initiated the lift off, the flight control system is trying to get the aircraft to a position which is continually moving because of the inaccurate GPS. this could and I’m sure has happened to others. I understand and respect Chris’ point of view. It is good to understand the system you are flying and that it is not perfect. You should always be ready to take control, meaning switch to a manual mode, to keep the aircraft from rolling over. I’m always cautious in flying using modes like loiter or even auto. Always standing by to take over in case the system does something unexpected. So I guess I would encourage use of these modes with the understanding that you need to be prepared to take control. But I totally understand your desire not to because the system is not completely trustable and these are very expensive machines.

Thanks Bill for your input, it happened so fast as it only lift off 15 cm before roll so i can only move sticks opposite end. We found out the servo was faulty but actually it was fine but the servo arm holder got loose which was plastic. This is not a new heli almost 4 years old and i am second owner so i guess that small plastic arm caused the problem.

I will do another test run shortly and see how these modes behave and be more cautious on taking over.

I noticed servo #2 went to full minimum value (and that servo is reversed), so it was trying to correct for it. But it was likely too far over already. The only way out in a dynamic rollover is to drop collective. Unfortunately, that’s not possible in Loiter because of the way the collective works - all you do is make a rate request to the autopilot with the collective - it is not full collective control.

You only have normal control response in Stabilize or Acro. Even a “GPS Glitch” (which is logged for any number of reasons for the GPS system) can cause a dynamic rollover like this one - I only recovered from it by switching to Acro right away and hunted around with the cyclic to find the attitude solution to get it back on the skids. Stabilize would’ve been a safer bet without having to hunt for the “ghost” but I didn’t have it set up for that at that time. So it ended up rocking back on the tail before I found it.


Just be aware GPS is most inaccurate close to the ground and these things can and do happen because of it. Helicopters are different than multicopter because they don’t just stop running and give up when you land or are taking off. A larger helicopter can easily take off again in spooldown, or roll itself over. That never happens when the pilot is flying it manually.

In flight where the GPS has a clear view of the satellite constellation with less chance of picking up multi-pathed signals, and the autopilot does not have to deal with the unique properties of helicopters close to the ground, the autopilot is great. We see virtually no failures of the autopilot system in flight. It’s most always the air-to-ground, or ground-to-air, transitions where the accidents will happen using an autopilot.

Your servo horn failure would’ve been easily recoverable in a manual flight mode and I see you did drop the collective. But it didn’t respond to it in time.

The other reason I normally only use manual takeoffs and landings is because many places I fly from there is no room for error. Sometimes I can use a road. But more often than not all I have is a field drive that’s overgrown with 4 foot tall grass. The main rotor is 6 feet across, so I stomp out a 8 foot circle to take off from, then on takeoff have to fly between a couple fence posts, under powerlines or around a couple trees and get lined up on the first waypoint. Coming back in is the reverse of the takeoff and it has to be precise within a few inches The autopilot simply can’t do that. A human pilot can.

OTOH, the autopilot is often fine in the critical flight stages in a wide-open area where there is little chance of picking up reflected GPS signals. And there is some room for error if the GPS position is not quite accurate. But if there’s trees, powerlines, buildings or other obstructions around never trust the autopilot to fly it. I’ve had flights where I got caught in a storm before and rain and wind wrecked my stomped out helipad before it got back, and I’ve landed it on the tonneau cover on my pickup with the main rotor only inches from the back of the cab in 15-20kt wind. You just don’t let an autopilot do that.

1 Like

Any chance you’re both right? GPS glitch, combined with the Loiter take-off bug (leash moves around) led to a dynamic rollover?

I must be mad.
Just did a mission today with my TR600 with all objects around you mentioned should not be there when flying auto… Manuel take off next to a big house on the lawn.
Trees everywhere, power lines further up on the hill. 300m away a massive overland power pylon about 30m below my mission.
The mission had 52 way points and was a smooth flying, including the RTL with auto land on the lawn.
I could proof it.
The Heli is flying in mode1 again with my new Castle ESC.
AC 3.6.0 rc11 ChibiOS, Pixhawk2.1 with a massive solid aluminum block underneath, normal HERE GPS unit.

Because I many times have to take off from field drives, there is usually powerlines directly overhead. I don’t consider it safe to use augmented flight modes when doing that.

In aviation there’s an old adage; There’s old pilots and there’s bold pilots. There are no old, bold pilots.

Just because it’s a RC “drone” I don’t believe in pushing your luck and blindly trusting automated systems.

I forgot to mention again that I never do a mission with my Helicopter without a test run with my small Quad-copter before hand.
I did so in this case with a clone Go Pro underneath the drone as a proof. This is the confirmation that the mission is working correctly and the sensors do not get dangerous interference.
Surely this way it is pretty safe to do than the Mission with my Helicopter afterwards. And it has been working for me each time!
Look at the photo out of the video. the arrows are power poles and the circle is the lawn for take off and landing.