Hi all,
I'm trying to figure out what caused a rather beautiful crash yesterday--if such a thing exists--of a custom built heavy-lift hexacopter that has had several weeks of problem free flight time. It's running a Pixhawk with ArduCopter 3.1.5 Hexa firmware. While the hexacopter was uncontrollable, its flight path was so mathematically lovely that I suspect there is a very simple explanation for this.
I'm specifically trying to determine which logs values can help diagnose this. I've attached the Archimedean spiral my hexa flew at full speed when I put it into "Auto." The onboard mission was a rather boring rectangular grid for sampling scientific data. I've flown it many times before. However, yesterday, my copter entered Auto like a bat out of hell and I couldn't regain control after switching to loiter, which was confirmed by telemetry.
The only control I seemed to have was throttle, so I quickly but gently brought it down in a field and fortunately damage was confined to a broken arm, several props, and the gimbal rail. In any event, I sometimes fly expensive instruments and having gone through my share of "start up" crashes over the years, this was unprecedented in my experience and really freaked me out.
Any help/advice would be deeply appreciated.
Replies
Here the screenshot of what I see when logged on droneshare.
Click on your name and you will get or be able to add an airframe and a log from there.
If you upload to Droneshare, it has auto analysis tools that will help diagnose. You can also link to that here, to make it easier for others to help.
PS, from a quick analysis on Droneshare, it looks like you've got terrible magnetic interference. Have you placed your mag very far from motor controllers, etc?
Thanks very much Chris. (I replied already but it doesn't seem to have shown up here so I'm writing again.) The uBlox is on a raised platform on the top board on a standard folding mount. I've attached a picture. The ESCs are far away underneath the motors. The only nearby things are 5.8GHz video, 915MHz standard 3DR telemetry, and 2.4GHz Jeti Duplex R18 with telemetry.
Given the GPS was working, could magnetic interference account for the bizarre flight path and extreme speed? Also, the system had been flying perfectly on Loiter before I switched to Auto--when it instantly went crazy--which would seem a very strange coincidence. In any event, I didn't think to switch to Stabilize as it was moving so fast I reacted without thinking.
Any thoughts are appreciated, as I can't afford to let this happen again, and I'm at a complete loss.
hexa.jpg
I appreciate Chris's response, but I'm trying to figure out how magnetic interference could account for a fly away like this, particularly as the GPS input seemed fine and Loiter mode was woking without difficultly immediately prior to entering Auto. I'd like to understand:
Many thanks,
Mike
Something is causing interference, be it radios or wiring. I suggest you upgrade to 3.2 (beta link in mission planner firmware) and recalibrate your compass. It will use both compasses (GPS and built-in Pixhawk) and give you more diagnostic data.
I suspect it's the forest of radio antennas you've got going around the GPS. You might try powering off the video and see if that eliminates the warning in the datalog analysis.
Thank you Chris. I suppose it's time to consider bottom mounting some of those antennas, but we have an Iris that displayed essentially identical behavior in the same location. I am guessing a large, nearby radio tower may be the cause of this intermittent interference.
Is there any already existing way to prevent this type of runaway behavior via parameter-triggered sanity checks? It's not clear to me currently what logical safeguards are currently in the 3.1 or 3.2 versions of the firmware. But preventing airspeed from increasing in response to magnetic interference seems like a reasonable step and could be GPS based.
I didn't think to attach the entire log to my initial message. My bad. Here it is.
2014-09-01 17-16-15.log