Skywalker 1900 crash during autotune

Hello all,

We have recently completed our build of a Skwalker 1900 airframe with the Params of our other perfectly working Skywalker. When testing the plane it responded well to commands on the ground and flew well in FBWA mode. We switched to Autotune mode and we feel like it flew ok for about 3 seconds then started flying very erratically in the roll and pitch axis. It then went out of control and we switched it back to FBWA although the plane was still not responding correctly, flipped upside down and plummeted into the ground.

How could a plane flying well in FBWA act so poorly in autotune? We are now to scared to use autotune again on our other airframe and will have to do it manually.

I have attached a download link for the log file for the crash and the previous flights. If anyone can provide any information on the cause of the crash that would be much appreciated

drive.google.com/file/d/0BzOTt- … sp=sharing

agis_ltd,

In looking at the binary log file, I noticed that the Skywalker was flying a 30 (mps?) or less when it was responding to the commanded roll. As you climbed to 100 meters (I think the units in the log are in meters) it looks like you had a throttle at about 75% but the speed still got up to over 40 mps. At which point the Skywalker was not rolling to commanded roll angles.

Perhaps your servos just could not handle the loads at those speeds. I noticed that when you flew a little slower that the Skywalker responded to the commanded roll angles.

Typically I fly my Skywalker at about 12 meter/sec. If you got the Skywalker flying at 40 meters per second it is some sort of record. :wink:

Hi there.

Thanks for the reply. The Skywalker was defiantly not flying at 30mps. We do not have an airspeed sensor so do you think the issue was with the GPS?

Thanks,

Tyler

Tyler,

Mission planner must have been plotting velocity in some strange units when I first reviewed your log. I reviewed the log again and I see that you only got up to about 20 m/sec. Still pretty fast for the Skywalker.

At higher speeds your control surface servos can stall, overheat or draw so much power that they overload your UBEC. For the initial tuning you should be flying at a nice easy 12 meters/sec.

With an airspeed sensor the controller will adjust the control gains for air speed, but without one you will have to take it easy.

If you have telemetry, flying with telemetry and a laptop to monitor your speed will allow you to get things under control.

Hi, I have seen similar behaviour in FBWA and AutoTune, it seems to be random. I put about 5 packs through my plane in AutoTune and FBWA. It had been stable. Made some manual adjustments to the pitch PID to back it off and flew again with out rebooting (because of the ESC bug). Crash was similar to your description. Glued it back together and went again. Log output attached from takeoff in FBWA. Plane was rolling left, RCin1 has right roll correction, NavRoll was commanding right roll, RCout1 (pilot, didn’t respond). This time it went through a barbed wire fence on knife edge. Sliced through the left wing root, carbon fiber spa and horizontal stabiliser. :astonished:

I can post the full log if it helps.

Were you running Mavlink with a connected GCS when it happened . I was using Droid Planner 2 / ULRS 1.6 downlink. With tower / ULRS 1.6 the plane was uncontrollable and I didn’t try to fly, it seems to overwhelm the Mavlink channel.

Here is the result