32" oscillations and crash in Loiter just after take off

[Mitar111 (Michael)]

It was first flight that day. I armed the drone in Loiter mode and raised the throttle. It climbed about a couple of meters and oscillations started with angle up to 50 degrees. Drone with 32" propellers was too close to me and I dropped it on ground without any loss.

That drone was tuned by me before and has total flight time about 3-5 hours. It was not ideally tuned for strong wind and high speeds, but it was flying very predictable before.

When looking into logs I don’t understand what caused the drone to pitch for 50 degrees.

Att: Roll vs DesRoll looks good. Therefore stabilize mode pid tuning is acceptable

PSC: VX vs TVX looks totally incorrect

RCIN: I was compensating pitch angle with stick. It was better to switch to stabilize, but it was too close and too fast. RC1 and RC2 are normalized (/100 -15)

ArduCopter: 4.0.3

Log file:2020-07-10 15-48-17.bin (844 KB)

Parameters:crash 10 07.param (18.2 KB)

[hosein_gh (Hosein Ghanbari)]

i think it was because of high vibration in all axis with clipping

high vibration will affect EKF measurement and could make problems like this

[Mitar111 (Michael)]

Thank you for reply

High vibration and clipping appeared later. Oscillations started with normal vibration level and without clipping. Most clipping and Vibe appear after landing.

If I check Pitch from different sources at matches very good. For me it doesn’t look like EKF measurement error. It least for Roll and Pitch.

But position and velocity estimate looks incorrect. How could it happen, that Pitch and Roll are correct, but velocity is incorrect? What could lead to VX != TVX?

[Aorish_Imtiaz]

Your desired angle also crossed 40 deg which is strange. The pitch controller just followed the desired pitch command. Haven’t you put any lean angle limit? I haven’t seen your log or parameters will do it over the weekend!

[Mitar111 (Michael)]

There was 40 or 45 degree limit. And it overshoot it for several degrees. I don’t think it is critical for such aggressive movement.

[xfacta (Shawn)]

32inch props and motors to spin them are not cheap - perhaps spend more time going over the Tuning guide
https://ardupilot.org/copter/docs/tuning-process-instructions.html
And use this spreadsheet to set a lot of your parameters

Initial parameters.xls

Put in your prop size and Lipo battery cells. I’d suggest some changes as listed below.

Vibrations are definitely something to be looked into.
Also the +5vdc supply to the flight controller looks like some device is drawing too much current, or you’re not using a standard power brick

Sort that out before you get a mid-air brown out and crash.

Check and change all these:
ACRO_YAW_P,0.8
ATC_ACCEL_P_MAX,13600
ATC_ACCEL_R_MAX,13600
ATC_ACCEL_Y_MAX,7200
ATC_RAT_PIT_FLTD,8
ATC_RAT_PIT_FLTT,8
ATC_RAT_RLL_FLTD,8
ATC_RAT_RLL_FLTT,8
ATC_RAT_YAW_FLTE,2
ATC_RAT_YAW_FLTT,8
ATC_THR_MIX_MAN,0.1
INS_ACCEL_FILTER,16
INS_GYRO_FILTER,16
MOT_BAT_VOLT_MAX,50.4
MOT_BAT_VOLT_MIN,39.6
MOT_THST_EXPO,0.81
MOT_THST_HOVER,0.2
BATT_ARM_VOLT,44.3
BATT_CRT_VOLT,42
BATT_LOW_VOLT,43.2
BATT_FS_CRT_ACT,1
BATT_FS_LOW_ACT,2

Also set BATT2 setting the same as BATT

Set this for smoother RC input:
ATC_INPUT_TC,0.22

PIDs look a bit agressive for such a large craft at this stage of tuning, probably try these:
ATC_RAT_PIT_P,0.11
ATC_RAT_PIT_I,0.11
ATC_RAT_PIT_D,0.005
ATC_RAT_RLL_P,0.11
ATC_RAT_RLL_I,0.11
ATC_RAT_RLL_D,0.005
After fixing the vibration and other issues, see if you can get it airborne and stable, hover for a while in AltHold, maybe try some gentle stick movements. Post a link to that .bin log.