Copter veers far off course in Auto mode -- help!

Hi all, we’re using a 3DRobotics RTF X8 copter with APM 2.6 (firmware v3.1), Mission Planner, and a Spektrum Dx7s controller. After many days (“Ready to Fly”???) we’ve finally managed to get our copter off the ground and flying fine in Stabilize mode. We tried Auto mode for the first time today and are having a new problem. The copter seems to be totally out of control when we flip it from Stabilize to Auto – it’s completely unpredictable which direction it will fly (though it’s often at 90 degrees to the programmed path). We tried two very simple flights, which involved three waypoints and a total travel distance of about 100m. The first time we ran autopilot, the general shape was correct but it cut the corners completely and never managed to hit the WPs. The next few times, the copter went off in completely the wrong direction and we had to retrieve it in Stabilize mode.

Attached are two logs. The 01-26 file is from our most recent flight, during which the copter veered far to the left almost immediately and then suffered a battery failsafe despite having around 80% power. The 01-13 is from our second most recent flight, during which the copter flew straight for about 10m and then veered far to the left (note that on an earlier flight, it had veered to the right). We managed to regain control and land it successfully. (We’re not super proficient yet at interpreting these files, so I BELIEVE these files correspond to those flights, although we think each of the logs might contain more than one flight.) We’re afraid this may be a compass issue but aren’t sure how to isolate it, or how to fix it.

I would very much appreciate any insight into this issue!

I suggest using ‘compassmot’ should be your first step

Hi rickp, thanks for the response. We’re using the compass on the APM. We just ran compassmot three times and got values of 13%, 19%, and 16%, which the APM website tells us is fine.

Any other thoughts on what could be going wrong?

What have got waypoint radius set to? If larger than you would like may explain the corner cutting.


At one time I had a problem that sounds similar. The quad wouldn’t flight straight segments on Auto missions. They were all arcing. Turns out I had the sign wrong in my magnetic declination value. Magnitude was OK, but the sign was wrong. Changed the sign and everything was fine after that.

The problem is that the COMPASS_ORIENT parameter has been set to 9. This should be either 0 if you’re using an APM with an onboard compass or 8 if you’re using an external compass.
I think the people setting up the RTF have made a mistake with this parameter and in fact, I found the same issue with an X8 that I ordered. I’ve contact 3dr and asked them to check their set-up procedure.

Hi all, thanks so much for your help. Unfortunately we had a catastrophic crash today in stabilize mode for no apparent reason. We tried all of the above mentioned compass fixes and nothing helped the autopilot, so we gave up on that and instead chose to fly manually, which has never been problematic in the past. Unfortunately, our drone suddenly started to drop and drift in AltHold mode, so we switched to stabilize. We had control for a few seconds and then we completely lost control of the copter, none of the failsafes kicked in, and despite all of our efforts to recover control, it crashed from 20 ft and broke more parts than we have spares for. I’m attaching the log.

Here’s a graph of RollIn vs. Roll:
[attachment=1]Screenshot (1) copy.jpg[/attachment]

And a graph of PitchIn vs. Pitch:
[attachment=0]Screenshot (2) copy.jpg[/attachment]

(both of these graphs look problematic to me). Here’s a link to a video of the incident:

[youtube] [/youtube]

(Unfortunately, the video cuts out milliseconds before the crash because our pilot requested that our cameraman stop filming).

We would very much appreciate any insight into what caused this. We have a feeling that our drone/APM may have had software problems when we received it. Has anyone had any luck getting tech support/a repair from 3DRobotics?

you have at least two serious issues here, one is most likely a motor, propeller or arm coming loose, at least tha data, that did not log commanded roll/pich looks like that, but we don’t really need that, because it’s stable initially.
Then you have the UAS operator, who clearly gets his priorities wrong, asking for stop filming in a critical situation. at the time I see stabilize, land, stabilize, land, and stabilize mode being switched on and off. Covering own ass should not be pilots first priority, if this was a a serious situation, you can compare it to a pilot, instead of handling situation, looking for, and pulling CVR fuse. The video is the closest you have a CVR, so once you make it during testing, everybody should understand it’s role, and not be concerned about own pride. That said, the few final seconds would not really help.

When repaired, you should check vibration leveles.

Andre – thanks for the response. How can you tell that it was a motor/propeller coming loose? We have had problems with prop nuts coming off, as they are not cross-threaded (which seems like a poor design). But how do you know this is happening? And if that’s the case, why does the drone seem to stabilize when our pilot switches to Land mode (as you noted), and then go haywire again when he’s in stabilize? (Note that he was switching between the two to attempt to regain manual control as we don’t have confidence in the ability of this copter to control itself).

I see the pitch/roll variation in the log increases quickly well after takeoff, and these changes increase in amplitude prior to the mode switching and crash. such change is usually linked to a physical “change” as something changes charasteristics due to getting loose.

Once you enter land mode, throttle is reduced to all motors until desired sink rate is achieved.
If you have vibrations, that could change frequency, if, like you mention; a propeller is slipping or loose, that would reduce slip and give the AP little less “unexpected” performance to work with.
Also, losing thrust on one motor on a hexa (or bigger) configuration would allow the other motors to compensate differently in different load scenarios: example, if you loose a propeller in an overpowered hexa or octo (nominal motor hover thrust=30% ), you might barely notice, while if a machine is less powered, or nominal thrust is 70% , the others will struggle much more to compansate. This due to the math of power duistribution, - remember than compensation for a lost motor does NOT know what motor is lost, and not only the two neighbouring motors will be affecte, the opposite motors must reduce output too, giving an asymmetric thrust distribution.

(Note that he was switching between the two to attempt to regain manual control as we don’t have confidence in the ability of this copter to control itself).[/quote]

Once you lose confidence in AP, your POH procedure should suggest to switch to Stabilize, and troubleshoot from there. Stabilize is the “least advanced” mode, with only a neccessary amount of programming involved. Once you switch to “Land”, barometric sensor(or sonar) is involved.
Stabilize let you “feel the machine” a bit better.
On any suspicion of bad vibrations and gyro drifting caused by them, land ASAP.

While I don’t see your choice of prop/mounts, threading is not really a bad problem.
You should tighten well, and when using CF propellers, also use anaerobic retainer between prop and base, and top plate, not only threads. Loctite 243 is fine.