Testing if the plane is armable before going to the airfield the following happend in my garden:
Arduplane was able to arm the motor (with check of gyro, GPS, compass, baro and RC). Motor was tested for approx. 1 sec, then the throttle was forced to 1000 msec in the transmitter (arm-switch).
Suddenly the yaapu told me that the fence was violated and the motor started to bring the plane back home…I have had troubles because arduplane was configured that a modeswitch is only possible within the fence. (fortunately the garden is too small for a complete takeoff )
The log told me what happened: “at the same time” RTL was triggered the GPS showed a rapid move of the plane of several 100 meters…
This happened some month ago - I did not report this because I found my personal solution of this problem: since then I configure a RC-channel to my arm-switch with “arm-emergency motor stop” (165) and never ever will use fence_option bit 0.
Maybe a RC channel with “arm-emergency motor stop” should be mandatory or at least default…
Unfortunately I lost the log because a windows update screwed up the GRUB and I have had to restore the system from an old backup.
Without a log file we can only make guesses to the problem.
Some possibility could be:
Some of the arming checks were disabled. If set correctly you shouldn’t be able arm until Ardupilot is satisfied with the position to prevent fence breaches and incorrect home positions.
There are general configuration errors.
Are you setting up some kind of throttle lock in the radio?
That is configurable in he RC switch options. Given the range of use of this firmware one persons mandatory is another’s unusable. That said I do use it on some of my planes.
Hi Allister,
Yes, the GPS was checked in the arming-procedure (I have had to wait some minutes to get a GPS-lock).
Yes, the throttle lock in the radio is in (forcing the throttle to 1000 msec, 0 %).
No, during this test the throttle lock was not on an RC channel.
The log showed a GPS mispositioning (straight line) of 428 m (I know that because the end of the line is very easy to remember - it was at the next school). The fence was configured at either 300 or 400m (my personal defaults).
What happened is very clear: after receiving a good GPS lock for arming the position got inaccurate due to a weak GPS-signal. This triggered a RTL and a start of the motor. The motor was stopped when the position was @home again (mode back to FBWA).
Forcing the throttle to 1000 msec will not prevent starting of the motor by RTL. I hope that the emergency motor stop has a higher priority than a motor start by RTL (I never tested that) .
Hannes
Hi,
isn’t it mandatory to have a break in the car? (or is it here also the wrong wording (I am german speaking)). Maybe: advice, hint, tip, recommandation??
But the plane was hitting my bush in the garden while beeing on the ground…(ok, I will stop this funny part of the discussion and step back to the topic).
As a elective you can do this. I have a VTOL that I can Arm or disarm right now using a RC. BTW you can use the Yaw arm to disarm it as well. And a arming switch! As you see making this a default is a bad idea. At times i have to disable arming checks to build. But after the build i will turn them on.
Was the parameter ARMING_CHECKS anything other than 1? What you are describing sounds to me like the EKF did not have a satisfactory home position prior to arming. Just because you see a gps position doesn’t mean that all systems are good yet.
With full arming checks and a geofence then you should be safe to take off in any mode because the home position will be a accurately set. It will take longer to arm the first time, but for the safety factor it’s worth it in my mind.
Emergency motor stop will kill the motor right away, but it won’t immediately disarm the drone. So be careful if you’re too quick on the switch it will get exciting again. The craft will need to sit still (as it would on the ground after a normal landing) for a few seconds before it will fully disarm.
Hi Brandon,
completly off topic but maybe from interest for you:
I am also flying a 3d printed plane (TA152) for 2 years up to now.
The arming-check I did also in the garden…waiting for the GPS-lock caused the softening of the surface of the wing in the sun at 34 degC (PLA, painted in yellow).
Hi Allister, for sure it was different than 1 - with 1 I would never be in the position to arm…the min. arming-check on my planes is Baro, GPS-lock, board-voltage (when equipped with speed-sensor also airspeed). But in fact: without GPS-lock no arming, without GPS-lock & arming no home-pos, without home-pos (and terrain-data) no fence, without fence no fence-triggered RTL - according to those facts there was a valid home-pos. And: I was curious why ardupilot thought the plane is “possibly in flight”: the RelHomeAlt was somewhere between 0 and 0.5m (starting with arming) - so I would say again, yes there was a valid home-pos.
AND: The GPS-pos was suddenly off the real position by 428 m on the map. So: homepos was Ok, GPS have had a mispositioning which triggered this.
The general wisdom shared here is that if you can’t arm with the arming checks set to 1 then you need to trouble shoot the reason why. Hardware that is not included like rangefinders, etc will be excluded so you don’t need to worry about that. You didn’t mention if you had the INS checks included, so it’s possible that the EKF did not have a satisfactory position. It is normal that you’ll have a solid GPS lock, but it takes some more time for the EKF to register a satisfactory position.
But again, without a log, this is all hypothetical and we won’t come up with a final answer. This car has no bakes.
Hi Allister, Reproducing the problem:
I reconfigured an up to now unused plane and tried to reproduce the problem:
yes, it was arming also with armcheck “all” - so I learned that there is no need to configure the hardware-setup manually.
yes, the car has breaks now (log available). (but for some reasons no propeller ).
yes, I was able to reproduce the problem (RTL triggert when the GPS signal is too weak).
(Motor started, no way to stop it from the radio without one RC-channel on option 165.)
I am going to reproduce the problem at the airfield and will send you the log - might take a while, right now it’s a little bit to windy - no fun flying always in the same direction…
Rough overview: first message: AHRS DCM active, then: Fence breached, RTL and after some (6) seconds later: EKF3 IMU0 stopped aiding.
To stop the motor I took away the metal cover I used to weaken the GPS-signal - having a GPS lock the motor stoppt beeing within the fence and throttle stick at 1000 msec.
When going to the airfield I will configure the Emergency Motor Stop.
Hi Allister,
here is the logfile from reproducing the problem:
17:32:03 first GPS-shading without an issue.
17:33:11.479 Motor started / RTL / second shading-test of the GPS
after this there is a normal flight in the log.
)
the issue is hitting the RC when it the air.