Discussions

Copter: 3.0.1: Near flyaway, not sure why

Posted by Rob G on July 18, 2013 at 6:50pm

Hi Guys, Need some help here...

One of the guys at our club had a near flyaway today - from his account his hex was in ALT_HOLD when he lost cyclic control and it started to fly away. Fortunately the area was surrounded by trees, one of which caught the runaway.

Attached is his dataflash log - seems the runaway started at approx 2715.

At this point ATT.RollIn goes to -7.8 and ATT.PitchIn to -12.2. The copter accelerates to approx 10m/s (GPS.Spd) before the tree steps in.

Need a more experienced set of eyes to help figure out why this happened.

At a loss - both my 3.0.x APM multi's have been working great and his should be a near copy of mine.

Logs attached.

DJI 550 framewheel / APM2.5 + MTech GPS + 3DR radio / 4S 5000

I'll ask Andrew to supply any details I may have left out.

Any help appreciated,

Thanks!

- Rob

2013-07-18 14-02 17.log

You need to be a member of diydrones to add comments!

Join diydrones

Email me when people reply –
Follow

Replies

  • Andrew Ellbogen July 21, 2013 at 5:55pm

    After this flight I added a new satellite and re flashed the eprom with a re calibration.

    I was about 2 feet off the ground gps locked and all seem fine activated a few waypoints and after the first there was something wrong. I hit RTL and it overshoot home by 60 feet. I tried to regain control but radio was locked out. I did not have a loss of signal  but the board locked me out. Bench test with a meter showed that there was still signal coming from the AR8000.Posted are some new logs.

    2013-07-21 13-56-57.tlog

    2013-07-21 13-58 22.log.gpx

    2013-07-21 13-58 22.kmz

  • Developer
    Randy July 19, 2013 at 10:56pm

    Looks like a radio failure to me.  If you plot the Pilot's input for roll, pitch, yaw and throttle that they stop updating at about 2700.  Generally I'd recommend setting up and testing the throttle failsafe (instructions here) so that in case of a radio failure the APM knows and responds by doing an RTL.  In 3.0.1 we also have the fence feature which if it were enabled would likely bring the copter back if it flew off too far.  Of course for these features to be used the pilot should first make sure that RTL works.

    3692778239?profile=originalBy the way, the compass offsets are a bit high.  I'm wondering has the pilot done the live compass calibration since upgrading to 3.0.1?

        X:-183, Y:143, Z:-145

  • Moderator
    Graham Dyer July 19, 2013 at 7:06am

    Just had a quick look and something that stood out for me from the log is that roll and pitch input seem to be way off for a period before the crash.

    3692777625?profile=originalAll is good until line 2730 then everything offsets by a substantial amount. This only happens to roll and pitch, yaw stays normal.

    There's also a quite a lot of vibration in X, Y and Z axis.

    That's all I can see at the moment, do you have a tlog?

This reply was deleted.

Activity