3.4.2-rc2 autotune query

Just tried to carry out an autotune on my quad (roll axis only). I tried this twice but ran out of battery before completion each time. Wondering if I can better set the starting parameters to give this a better chance to finish quicker. My quad:

Turnigy Talon v2 frame cut down to 330mm
2814 kv870 Sunnysky motors
4s 2600mAh 65c Turnigy A spec pack
8045 carbon/nylon props - very stiff
20a X-Rotor ESCs
AUAV-X2 FC

MOT_THST_HOVER = 0.26
AUTOTUNE_AGGR = 0.065

Want to get a better tune as in even slight wind gusts it is far too twitchy. Today there was zero wind but snow on the ground so I imagine battery not running at peak performance compared to on an average day here.

Any advice?
Log file here: https://dl.dropboxusercontent.com/u/12131631/forum-links(do%20not%20delete)/ATUN1.BIN

Thanks, Paul

For a while now I have found it necessary to auto tune one axis at a time.

Just set AUTOTUNE_AXES: Autotune axis bitmask to one axis, do the tune, then set it to the next.

The Wiki page

I was trying roll axis only. Thanks

Hi Paul,

This is a very strange. I have not seen anything like this before.

So the good:
Vibes
motor output match
tests seem to be nice a regular suggesting the weather was nice

The bad:
P had to drop ridiculously low to stop overshoot
There appears to be a significant 50Hz oscillation going on
You haven’t set the MOT_BAT_VOLT_MAX and MIN (they should be 16.8 and 14)

The consistent overshoot is the funny one. How have you set up your ESC’s?

Can you provide some photoes, particularly in the autopilot mount, and have a look for any movement in the frame.

Leonard, Apologies, I didn’t pick up on your reply sooner. My copter can be seen in this gallery: https://goo.gl/photos/g4BwUr8mF75CSkUBA
Check the last few pics to see how it is now and some build details. Thanks as always, Paul

Just wondering if the 50hz oscillation could be caused by the servo I have for camera tilt? Only thing I can think of that might run at 50hz.

I forgot to add, the ESCs are completely stock X-rotor 20a. No special config at all aside from throttle calibration.

Hi Paul,

So the frame should be very stiff. Props and motors should react fast. So that leaves two possibilities vibration mounting and ESC firmware.

While I don’t think it is the servo it might be a good idea to leave it unpowered while you do the autotune.

Could you strap the autopilot down so the vibration dampening is effectively removed and run another autotune. If that doesn’t change anything then the only other thing to do is to flash BLHeli onto the esc’s.

Is there anything else on that frame that can bend and wobble? The gps mount for example??

Leonard,
Yes, the frame is totally solid and the props are also - hard to bend them. The motors have no play in the bells also, and the GPS mount is a solid aluminium pole - super rigid. I can certainly try tying down the autopilot - as you can see in the photos, it uses just a double thickness of sticky pads, so is already fairly firm in its mounting. If I can please ask - what are you looking at in the logs to determine that 50hz wobble? I looked at the gyro and accelerometer outputs and didn’t see anything specifically at 50hz.

Thanks again, Paul

If I may ask, why 3.4.2-rc2 when 3.4.2 is released?
That might clear up some stuff.

It wasn’t released on Nov 9th when I raised this :slight_smile:
Haven’t really had decent enough weather to try since (with 3.4.2 release).

I have actually determined that my short flight time could be down to my (Turnigy A-Spec 2600mAh 4s 65c) LiPo being a bit worn out - despite it being only just over a year old. Checked the IR and it is 75mOhms - so think its not lasted well. Going to try the next autotune using a better LiPo.