Pixhawk Hex Build -- Need Help Tuning

ArduCopter Copter 3.1
1
  • Pixhawk controller
  • Running APM 3.1.2
  • Castle Creations Fixed Endpoint ESCs
  • ESCs properly calibrated

Test fixture is a table with four strings attached to each arm and threaded through the table. At the end of each string is a 51g spool of wire. So the vehicle is effectively bound in a fairly small area and if I ever get into trouble, I go to max throttle which maxes-out every string.

Here’s what I see:

  1. Copter comes off ground cleanly (doesn’t try to flip as if props were rotating incorrectly)

  2. But does begin an almost immediate drift once it comes off the ground. If it weren’t in the test fixture, it would just take off laterally

  3. It has an almost violent side-to-side instability

  4. To land it back on the fixture, I basically have to catch it on a favorable side-to-side oscillation and just cut power so it flops back down

Please watch the video here:

dropbox.com/s/v0y97yxgdnwl6 … 135936.mp4

What do I need to do differently? Is this a tuning problem?

2

Another video (but basically the same thing):

dropbox.com/s/8hulrmp14we9r … 135847.mp4

3

Hi bcdebusk,

I think this is not a hardware issue, I’m moving this post to the 3.1 sub-forum.

4

Looks to me like you should take it outside where you have GPS lock and are flying outside of ground effect. Then do an autotune

5

I wanted to autotune, but it wasn’t flyable before.

I installed the PPM link, moved it to my DX9, redid all the calibrations and now it seems flyable.

I’ll try to autotune it once the weather permits.

Thanks all!

6

So I watched a youtube video on tuning in a fixture and got very reasonable P & D values. As the video recommended, I left I and 0 and “headed outside!” (that’s what the tutorial said).

I armed the system in stabilized mode, popped it off the ground and it immediately started drifting right into a fence and broke $76 worth of carbon fiber propellers.

I even duplicated the fixture from the tuning video and experienced all the things it pointed-out (i.e. tremors with values of D too high, sluggishness with P values too low).

But once I headed outside, it’s like the quad had a mind entirely of it’s own.