Autotakeoff failed to increase pitch - why?

descending from start, an manual input temporarily rectified the problem, but then the plane continued down…

dropbox.com/s/gzxzbjzxdursqf6/27.BIN?dl=0

Did you specify the angle of climb in the TAKE_OFF step of the mission ?
(I made the mistake last week : target altitude did not make it climb !)

yes, 8 degree minimum pitch up is set. (and see in both as cmd_param , and as desired pitch)
I wish a dev could look into it, even reported a bug as I am lost as to why this is happening.

Could you post your mission file ?

here is the simple test-mission:

QGC WPL 110 0 1 0 16 0 0 0 0 0 0 0 1 1 0 3 22 8 0 0 0 69.6461639404296875 18.792327880859375 50 1 2 0 3 16 0 0 0 0 69.6472549438476562 18.7855911254882812 100 1 3 0 3 16 0 0 0 0 69.6490859985351562 18.7869243621826172 100 1 4 0 3 16 0 0 0 0 69.6485671997070312 18.7942619323730469 100 1 5 0 0 177 2 10 0 0 0 0 0 1 6 0 3 16 0 0 0 0 69.6468048095703125 18.7936172485351562 100 1

Looking at your log and mission, The controler know that he must climb, and did climb to 9 meter then fall down… It is quite strange.
I can’t help you more, sorry and this forum doesn’t seem to be very active :frowning:

no - I did not assume you could help, I am used to reading logs as well, but this behaviour does not make any sense to me.
I created a bug report, I just think it’s strange noone else seems to have tried/discovered it.

Thanks for trying :slight_smile:

The mission you detailed above is not what was loaded in your plane. I can’t attach a file for some reason so here is the correct mission pasted below:
QGC WPL 110
0 0 3 16 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 1
1 0 3 22 8.000000 0.000000 0.000000 0.000000 69.646164 18.792328 50.000000 1
2 0 3 16 0.000000 0.000000 0.000000 0.000000 69.646339 18.786278 100.000000 1
3 0 3 16 0.000000 0.000000 0.000000 0.000000 69.646759 18.798164 100.000000 1
4 0 3 16 0.000000 0.000000 0.000000 0.000000 69.648071 18.780272 100.000000 1
5 0 3 16 0.000000 0.000000 0.000000 0.000000 69.648834 18.780785 100.000000 1
6 0 3 16 0.000000 0.000000 0.000000 0.000000 69.647415 18.798466 100.000000 1
7 0 3 16 0.000000 0.000000 0.000000 0.000000 69.647697 18.800827 100.000000 1
8 0 3 16 0.000000 0.000000 0.000000 0.000000 69.649384 18.779240 100.000000 1
9 0 3 16 0.000000 0.000000 0.000000 0.000000 69.649643 18.781258 100.000000 1
10 0 3 16 0.000000 0.000000 0.000000 0.000000 69.648300 18.798723 100.000000 1
11 0 3 16 0.000000 0.000000 0.000000 0.000000 69.645866 18.793961 100.000000 1
12 0 3 16 0.000000 0.000000 0.000000 0.000000 69.648331 18.791557 100.000000 1

I flew your mission in SITL and it flew fine including the take off.
I can see that you lost altitude on takeoff. The obvious question but are the control surfaces behaving the correct way around. I see you have aileron’s reversed. When you put the plane in FBWA and move it around do the control surfaces move in the direction you expect?
Thanks, Grant.

Hi.
gmorph : I am very glad seeing you taking a look into this.
Yes, the control surfaces were fine, like the flights before. And yes, it did move in correct directions.

Also, please notice RCout for elevator, you’ll see that it does not really even try. (expect for when helped by manual input) - so we can’t suspect an mechanical/servo failure.

It’s hard to explain rcout2 (for elevator) - and it proves something went very bad…

Also, if elevator were inverted, it should have pitched down at an increasing rate as deviation grew, but instead it looks like it just flew , barely doing anything at all about pitch.

If you flew it with master, is there any chance this bug somehow influenced anything ? github.com/diydrones/ardupilot/ … eeb2ac3461

This is a graph of the output channels with altitude. I assume the takeoff begins when the throttle zooms up to full (the orangish line)? If so it looks like its pitching down immediately. Was it a hand launch?
Thanks, Grant.

yes, a hand launch, autothrottle.

On a second review, it seems to me that RCOUT.ch2 tries increase, but extremely slowly - regardless of pitch difference from des.pitch

In the moment of launch, you can see RCOU.Ch2 go all the way to 1643 when pitch was -6.9 (line~23630) , but then it drops rcou.Ch2 back to 1531 already when pitch is raised to -2 degrees, - so it lets the servo output drop off as soon it corrected the negative pitch after launch, instead of pushing on for higher PWM.