This is a discussion re the ArduCopter-3.1 AutoTune feature.  A place for people to report successes and failures, ask questions and get answers.

The AutoTune wiki page is here.

 

Views: 74326

Reply to This

Replies to This Discussion

I'm anxious to have mine, I follow your autotune results :)

Ok, I finally got around to reassembling the Alien 680 with the BLHeli flashed Quattro ESC. Observations:

1) BLHeli is awesome. During Auto Tune I could see and hear the difference in motor response.

2) Auto Tune process: much quicker recovery, snappy, less altitude loss.

3) After Auto Tune: much better handling, copter doesn't overshoot like before.

Now for the bad:

1) In Pos Hold, it is still not stable in wind.

2) Descents are better, but still wobbles on occasion; gimbal cannot compensate well enough.

3) With APC MR props, vibration is high. With cheapo CF, much better, but I don't like cheap props as IMO over 400-500g thrust they flex at the hub (center) outward and cause oscillations.

4) PID's for pitch and roll are very diverged.

Conclusion:

Today I did a thorough inspection of the frame. At 3179g with GP and gimbal, it is just too heavy for this frame I think. The motor mounts are weak and it is easy to flex them. IMO, the frame is crap.  There are ways to strengthen it, but then you add weight. There are stronger motor mounts, but again adds weight plus expense. That doesn't fix the twisting of the arms, and with the props "thwopping" the frame from air wash I don't think helps either. I know others with this frame that are in ad continuum trying to remove vibrations and such. Overall I think the frame flexes too much and will likely disassemble and mothball it.

I will be looking for a strong frame, smaller in size and light after all equipment is attached. My eye is on these two. The XP2 is small, but the KDE 515 motors on 4s will run it with 12" props I'm fairly certain.  The XPX is "heavy lift", so not sure I want that; already have a 750 Y6 boat anchor. They are all aluminum, which may seem obsolete by today's standards, but simple in design and look to be strong. This is purely for using a GP and gimbal to take some decent video.

http://xproheli.com/collections/frames/products/xpx-quadcopter-frame

http://xproheli.com/collections/frames/products/xp2-quadcopter-frame

It looks to me they are having a fire sale.

Note: eCalc was WAY off on throttle %(log) on the Alien 680 @3179g. It predicted 58% throttle, but actual was 50%. I had this problem earlier this year and contacted eCalc about it. They changed it, but it's never been as accurate as it was before. I'm currently using 14" props. If 13" is plugged in eCalc predicts 74%, which I'd bet wouldn't hit 60% as at 2.7kg it was 51%. Just sayin" They are releasing a new version so maybe it will be better.

I have just done my 1st autotune (Pitch & Roll) on my S500 frame, 11x45 props and SS 3108-720 motors. 3S Nanotech 4500. ZTW Spider 30 ESCs. Weight 1450 grams.

The new PIDs were stored but I have no idea if they are "reasonable". They certainly have changed :)

I initiated Autotune while in AltHold mode and did a few manual corrections to avoid the trees :) Flight took 12 minutes until I heard a sound from the Pixhawk which I took to mean end of tune. A few seconds later I got a battery failsafe and the quad did a nice RTL for me.

Why would the P and I values be the same?

Why would the Stabilize Pitch & Roll values be 2-3 times higher than the default values?

Attachments:

Every frame is built differently, so the PID values could increase or decrease.

When the autotune is engaged, it tries to find a tune that works best for your frame.

Sometimes, you will find that, even after an autotune, you still have to adjust the PIDs manually for optimum performance.

I have a new build: a Robocat 270 with a Pixracer (AUAV) but I can't do autotune works, finally I try to manual tune, no wind and looks nice, cool Alt Hold like near Loiter, I try first with one axis (1) and nothing, only very short movements and then nothing, then I try with all axes (7) but the same thing, I controlled CG (not very easy on this little guy) mid thr is near 48 and althld works nice with stick on the middle; I'm shure that I missing some silly thing but I'm not figure what, someone can help me please?

Here my log:

https://www.dropbox.com/s/xzrmplqv6fcxs0e/failed%20autotune.BIN?dl=0

Hi Cala,

The problem here is your rate noise. Could you please try something for me. Could you set

INS_ACCEL_FILTER to 10

ATC_RAT_RLL_FILT to 20
ATC_RAT_PIT_FILT to 20

ATC_RAT_YAW_FILT to 2.5

LOG_BITMASK to 196605

Be a little careful on your first take off as the filter changes may affect you tune.

Thanks!!

Nice, thank's Leonard to take a look, I try like this.

Hi Leonard,

I am also having some problems with autotune on a new overpowered copter and posted a log as well as some details at:

http://discuss.ardupilot.org/t/autotune-failed-pixracer-overpowered...

Since the vibrations are low on my copter I think it is not the best idea to lower the frequencies. Anyway one question what do you mean with rate noise?

Thanks,

Thorsten

As posted at discuss.ardupilot.org I got a working pitch autotune after some tries. Too much wind now to run some more test...  

I try today, no wind, with this params but only change the copter become little movements more nervous

https://www.dropbox.com/s/w9sbq4cebukovpl/5%2016-06-2016%2011-17-02...

I am wondering what has the stronger negative influence on autotune - wind or some pilot input from time to time to bring the copter back. Not sure if possible but I guess it would help if autotune waits for 1 or 2 seconds after the pilot released the sticks because the copter is sometimes still drifting when autotune proceeds.

Hi Thorsten,

Yes, that can be a problem. My next major change to autotune will be to let it get started from Loiter. It would then hold position on it's own.

Reply to Discussion

RSS

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service