Quadcopter tilts in wind in level mode

[KasparsL]

Last days i have got time and good weather for flying and I am trying to eliminate all problems with hardware so I can upgrade flying skills without crashing because of hw issues.

I have noticed that my quadcopter in level mode sometimes tend not to recover if wind blows on it. For example - forward flight, release sticks, and it stays tilted - away from wind, so I need to counter react to stabilise it. It looks like wind tries to flip the quadcopter. Or if hoowering - wind blows and quadcopter tilts away from wind , and not just slides away.

Because of this i have almoust crashed a couple of times - I do some manuvers, release the Roll-Pitch stick, but quadcopter doesnt fully level itself( it can stay 30deg or more tilted), and I need to think fast which way to correct it.

The question is - is it because of PID tuning, is it because of vibrations, is it because of some callibration, is it a bug, is it just normal?

In flight and hoovering quadcopter is quite stable, only if descending fast >5ms, it wobbles a little.

short video of hoovering in Position Hold http://youtu.be/L55f9pwv4kY

[QuadBow]

Honestly, I don't see any big issue with the stability of your copter presented in the video.
If the flight control can't compensate a tilt of 30° in angle or level mode there are several reasons to investigate
- the motors are to weak
- the wind is too strong
- level P and I are too low

In my view, the angle / level implementation is quite dependent of the copter. Vibrations or missing calibration would impact the stability more or less always - not only in the described cases. For me it sounds like a mismatch of the level PID values.

[KasparsL]

If the motors would be too weak, would I be able to compensate manually? I don't think so. I am able to tilt it 45ded while hoovering and it will return back if stick is released. Maybe the PIDs are not sharp enough? I need to play with PIDs more I think... Ill try to record the flight next time, and maybe I will be able to see from osd, what really happens then.

[RustyPaint]

Sounds like PID tuning to me. Had the same problem until I upped the P value for ROLL and PITCH. Wind was enough to flip it, but now (after change in P), it just gets pushed around and remains flat.

[KasparsL]

Just had a serious crash because of tilting! http://youtu.be/pwfPgkLNW2Y Ad it is not because of wind. It just tilted! I was flying in horizon mode. After this I have upgraded to dev1317, and was practising flying in acro mode. And actually saw sudden tilting to front (20deg) in acro mode. This was small enough (and I wasnt flying fast) to cunter react.
I can give any additional info of my setup, just help me to get rid of this tilting on its own!

[dougroundup]

Maybe calibrate ACC ?

[KasparsL]

Maybe calibrate ACC ?

Yes I have calibrated ACC and also end points.
I you have noticed in video it is completley level first 1:50 min (flying in Horizon mode). And actually this tilting can happen also after 10min.

[KasparsL]

I tried to draw how it happens (my stick inputs on the bottom of the image mode2) :



Interesting thing is that quadcopter "knows" that it is not level (can see on OSD)! Tilting usually happens after a while quad is been powered on ( it is been flying several times, was armed disarmed several times ).

Here is a video with the tilting issue on dev1317. I found out it only when was watching the crash video. This crash was light and nothing was broken.
http://youtu.be/fMGPEEWFJj4

[copterrichie]

From what I observed in the video, appears more as the copter do not like descending in its prop wash. All copters have this problem but depending upon the PID settings, this can become a serious problem as you have discovered. IMO, each copter's PID settings are unique, but I am curious what your are at present?

[KasparsL]

In the last video, as I remember "P" for roll and pitch was increased to 5.5 and "D" to 30. and "P" for Level was increased to 8. Everything else left default. ( I have tried "harder" pid settings, quad is stable even with P=7 D=30 with no shaking/oscillating at all). I have defined MPU6050_LPF_98HZ if it helps. Quad is stable also without it (when props are balanced well - i do balance hubs also).

But if it is deal with prop wash, why I can descend at 6m/s straight down with a little shaking? But when performing this maneuver, there are no shaking what so ever and quad knows its level , it just tilts, like there is some other stick input? And I would agree if this would happen every time (at least every 3rd time), but no - its once in 5 - 10 min.

[copterrichie]

Hmmm, personally, I have always used the default when it comes to the MPU6050_LPF filters, I never had a problem with vibration in that aspect. The things that concerns me is, the D=30, that I believe slows the reaction time of the P value. I have never personally used anything over 23 but again, my copters are heavier than average.

[copterrichie]

P.S. I should also mention that I have had that descending issue on rare occasions with previous versions of the firmware. So I don't believe it is something related to the current developments.

[KasparsL]

Hmmm, personally, I have always used the default when it comes to the MPU6050_LPF filters, I never had a problem with vibration in that aspect. The things that concerns me is, the D=30, that I believe slows the reaction time of the P value. I have never personally used anything over 23 but again, my copters are heavier than average.

I think that PID values are OK.

How to better explain, when it happens, it looks like there is additional stick input to front, no oscillation or vibration more than usually. I have had problems with vibrations and oscillations before, caused by unbalanced props or wrong PID settings, but this is completely different.

[vpb]

I face the same annoying effect with horizon mode when flying with the wind, especially fly agaist the wind.

[tovrin]

I dont have OSD, so maybe I'm understanding this wrong, but the horizon shown on the screen appears backwards on roll.

[copterrichie]

I THINK I may have repeated your bug. It only happens under very rare Pitch down conditions. I have been able to repeated this on several occasions now and hope that I can recreated it on the bench. I lost track of which development version I am using, I believe it is something like R1277 but I am going to upgrade to Pre2.2 and test again.

[copterrichie]

I should also note that I am using a MPU6050

[copterrichie]

I am not sure if this is the problem but as I rotate the Pitch axis on my test board with the Roll axis fixed where it can not move, there appears to be an error introduced on the Roll axis, when the Pitch angle is greater than 45 degrees or there about. This error increase to 90 degrees. Here is a screenshot.

[copterrichie]

I performed the same bench test, this time using an ITG3205 Gyro and a MMA7455 Acc and the problem does not exist. Same version of the firmware.

[copterrichie]

I just confirmed the problem with is the MPU6050. I loaded the latest version R1350 and reran the test on both the MPU6050 and the ITG3205/MMA7455.

[copterrichie]

Erred..

[Hamburger]

grasp for understanding of this fix moved over to viewtopic.php?f=8&t=3127&p=31752#p31752

[KasparsL]

Hi,
I was away for some time, I red thru some posts about possible fix for my or similar issue, but still - is it confirmed /fixed ? Is it included in some dev version?

I have landed on a tree last weekend (7m above the ground - climbed up to get it!), because of this tilting, only one prop broken. It is anoying that I crash not because of my piloting skills, but because of some glitch - it is like crashing because of cheap prop snapping in the air by it self .

[copterrichie]

Please ignore that fix.