INav 6 horizon drift issue

[mykls]

I was testing INAV 6 on an AR Wing Pro and late in the flight (30 mins in) I had a horizon drift issue on stabilised mode (I usually fly on Acro), which caused an uncommanded left roll and later resulted in a crash when I accidently tripped RTH too close to the ground. I also test flew the INAV 6 build on some other planes and they performed fine. Blackbox log available and attached (zip).

DIFF ALL config below:

INAV/MATEKF405SE 6.0.0 Oct 20 2022 / 15:20:51 (d1f6f19a)

GCC-10.2.1 20201103 (release)

start the command batch

batch start

reset configuration to default settings

defaults noreboot

resources

mixer

mmix reset

mmix 0 1.000 0.000 0.000 0.000 mmix 1 1.000 0.000 0.000 0.000

servo mixer

smix reset

smix 0 3 0 50 0 -1 smix 1 3 1 50 0 -1 smix 2 4 0 -50 0 -1 smix 3 4 1 50 0 -1

servo

servo 3 1000 2000 1570 -90 servo 4 1000 2000 1533 115

safehome

feature

feature MOTOR_STOP feature GPS feature BLACKBOX feature PWM_OUTPUT_ENABLE feature FW_LAUNCH feature FW_AUTOTRIM

beeper

blackbox

blackbox -NAV_ACC blackbox NAV_POS blackbox NAV_PID blackbox MAG blackbox ACC blackbox ATTI blackbox RC_DATA blackbox RC_COMMAND blackbox MOTORS blackbox -GYRO_RAW blackbox -PEAKS_R blackbox -PEAKS_P blackbox -PEAKS_Y

map

serial

serial 0 64 115200 115200 0 115200 serial 1 0 115200 115200 0 115200 serial 2 2 115200 115200 0 115200 serial 5 2097152 115200 115200 0 115200 serial 30 0 115200 115200 0 115200

led

color

mode_color

aux

aux 0 0 0 1200 2100 aux 1 1 2 1700 2100 aux 2 12 1 1700 2100 aux 3 53 1 1300 1700 aux 4 11 3 1700 2100 aux 5 10 3 1300 1700 aux 6 57 4 1300 2100 aux 7 13 5 1600 2100 aux 8 26 0 1200 2100 aux 9 27 2 900 1200

adjrange

rxrange

temp_sensor

wp

wp 0 invalid

osd_layout

osd_layout 0 0 1 14 V osd_layout 0 1 24 12 H osd_layout 0 7 12 15 V osd_layout 0 8 13 14 V osd_layout 0 11 1 11 V osd_layout 0 12 0 12 V osd_layout 0 13 0 7 V osd_layout 0 14 1 1 V osd_layout 0 15 0 8 V osd_layout 0 20 17 0 V osd_layout 0 21 4 0 V osd_layout 0 22 15 0 V osd_layout 0 23 1 9 V

logic

global vars

programmable pid controllers

master

set gyro_main_lpf_hz = 25 set dynamic_gyro_notch_q = 250 set dynamic_gyro_notch_min_hz = 30 set gyro_zero_x = -9 set gyro_zero_y = -22 set gyro_zero_z = -16 set ins_gravity_cmss = 977.286 set acc_hardware = MPU6000 set acc_lpf_hz = 102 set acczero_x = 72 set acczero_y = -20 set acczero_z = -276 set accgain_x = 4080 set accgain_y = 4094 set accgain_z = 4044 set align_mag = CW270FLIP set mag_hardware = NONE set baro_hardware = DPS310 set rssi_channel = 15 set serialrx_provider = CRSF set blackbox_rate_denom = 2 set motor_pwm_protocol = STANDARD set failsafe_delay = 10 set failsafe_procedure = RTH set current_meter_scale = 256 set model_preview_type = 8 set small_angle = 180 set imu_acc_ignore_slope = 10 set imu_inertia_comp_method = ADAPTIVE set applied_defaults = 3 set gps_provider = UBLOX7 set deadband = 4 set yaw_deadband = 25 set airmode_type = STICK_CENTER_ONCE set inav_w_z_baro_p = 0.350 set nav_wp_radius = 1500 set nav_manual_climb_rate = 500 set nav_rth_climb_first_stage_mode = EXTRA set nav_rth_climb_first_stage_altitude = 4000 set nav_rth_climb_ignore_emerg = ON set nav_rth_allow_landing = FS_ONLY set nav_rth_alt_control_override = ON set nav_rth_abort_threshold = 60000 set nav_rth_altitude = 7000 set safehome_usage_mode = RTH_FS set nav_fw_bank_angle = 45 set nav_fw_climb_angle = 25 set nav_fw_dive_angle = 20 set nav_fw_loiter_radius = 6000 set nav_fw_control_smoothness = 4 set nav_fw_launch_velocity = 150 set nav_fw_launch_max_angle = 180 set nav_fw_launch_timeout = 10000 set nav_fw_launch_max_altitude = 10000 set nav_fw_launch_climb_angle = 19 set i2c_speed = 800KHZ set name = :WTED

profile

profile 1

set fw_i_pitch = 12 set fw_ff_pitch = 80 set fw_i_roll = 10 set fw_ff_roll = 40 set fw_p_yaw = 0 set fw_i_yaw = 0 set fw_ff_yaw = 0 set max_angle_inclination_rll = 450 set dterm_lpf_hz = 10 set fw_turn_assist_pitch_gain = 0.200 set heading_hold_rate_limit = 101 set nav_fw_pos_z_p = 35 set nav_fw_pos_xy_p = 70 set fw_level_pitch_trim = 5.000 set rc_expo = 30 set rc_yaw_expo = 30 set roll_rate = 34 set pitch_rate = 10 set yaw_rate = 1

profile

profile 2

set fw_p_pitch = 2 set fw_i_pitch = 12 set fw_ff_pitch = 80 set fw_p_roll = 1 set fw_i_roll = 10 set fw_ff_roll = 30 set fw_p_yaw = 0 set fw_i_yaw = 0 set fw_ff_yaw = 0 set max_angle_inclination_rll = 450 set fw_level_pitch_trim = 5.000 set roll_rate = 36 set pitch_rate = 12 set yaw_rate = 1

profile

profile 3

set max_angle_inclination_rll = 450 set fw_level_pitch_trim = 5.000

battery_profile

battery_profile 1

set bat_cells = 4 set vbat_min_cell_voltage = 275 set vbat_warning_cell_voltage = 325 set battery_capacity = 7000 set battery_capacity_warning = 1400 set battery_capacity_critical = 700 set throttle_idle = 5.000 set nav_fw_cruise_thr = 1550 set nav_fw_max_thr = 1800 set nav_fw_launch_thr = 1800

battery_profile

battery_profile 2

battery_profile

battery_profile 3

restore original profile selection

profile 1 battery_profile 1

---

• FC Board name and vendor: MATEKF405SE
• INAV version string: INAV/MATEKF405SE 6.0.0 Oct 20 2022 / 15:20:51 (d1f6f19a) LOG00017.zip

[breadoven]

Looks like the Baro altitude was decreasing during the final RTH Climb phase in spite of the pitch Attitude indicating nose up. Also servos 3 & 4 appear to be pitching down. This is based on their positions during the launch climb where 3 is low and 4 high. At the end of the flight during the RTH climb both servos have higher values with 3 higher than 4 indicating pitch down. So a large amount of horizon drift on the pitch axis. Is there no OSD DVR available to confirm this ?

Also why did you use Failsafe mode to trigger RTH then RTH mode itself afterwards ?

[b14ckyy]

The log also shows a MASSIVE amount of vibration induced in your Flight controller. In the 15-20Hz range and also in the 45-55Hz range the noise is critical. Accellerometer Z axis fluctuates between -5G and +5G at high frequency.

this could have completely thrown off the IMU. The Vibration amplitude raises with throttle but frequency lowers at higher throttle. That's a bit weird. But there is some issue with physical vibrations for sure.

[mykls]

Looks like the Baro altitude was decreasing during the final RTH Climb phase in spite of the pitch Attitude indicating nose up. Also servos 3 & 4 appear to be pitching down. This is based on their positions during the launch climb where 3 is low and 4 high. At the end of the flight during the RTH climb both servos have higher values with 3 higher than 4 indicating pitch down. So a large amount of horizon drift on the pitch axis. Is there no OSD DVR available to confirm this ?

Also why did you use Failsafe mode to trigger RTH then RTH mode itself afterwards ?

No OSD as I was flying with the DJI OSD. I'll extract some of the video for the last minute of the flight and upload if I can reduce the file size enough. The triggering of FS was accidental; I have a three position switch that has FS (used for two switch arm), Acro and Angle and when I switched to Angle it pitched down and rolled, so in my haste to push it back to Acro, I pushed it too far and it went in to FS. As the ground was approaching rapidly I was fumbling with the switches (RTH next switch over) to regain control, but got the wrong switch.

https://user-images.githubusercontent.com/58650613/201492789-82cd065e-7c1c-47b4-8709-4fef8333825e.mp4

[mykls]

The log also shows a MASSIVE amount of vibration induced in your Flight controller. In the 15-20Hz range and also in the 45-55Hz range the noise is critical. Accellerometer Z axis fluctuates between -5G and +5G at high frequency.

this could have completely thrown off the IMU. The Vibration amplitude raises with throttle but frequency lowers at higher throttle. That's a bit weird. But there is some issue with physical vibrations for sure.

The video I posted in my last reply shows how bumpy the last part of the flight was. Much more than I would expect given the conditions and at the start of the flight it was very smooth, so it's likely that the vibration started during the flight. My low-ESR capacitor on the flight controller separated in the crash, so it may have come loose in flight, which may have made a difference.

This was my second flight on INAV 6 for this aircraft. Earlier flight seemed fine and no issues on the previous INAV 3 flights.

[mykls]

Just checked previous logs and while previous flights showed a small peak of accelerometer noise around 170-180Hz, it was very small compared to this flight. Not sure what changed on this flight.

Has there been much change in accelerometer filtering between INAV 3 and 6? Two flights prior I was on INAV 3 and the accelerometer data looks good. The second last flight on INAV 6 showed a little more noise, but not much and this last flight, as you point out, was terrible, so I expect that it was a worsening mechanical issue, but I'm interested to know if the filtering has also been changed with version changes?

[b14ckyy]

@mykls one explanation could be that during the previous landing, the prop got a tip chopped off during landing and you did not notice? The vibrations are too strong and too regular to be a lose part inside the plane. But they are the most likely explanation. The gyro data look fine. so its only mechanical, not electrical.

[breadoven]

Hard to tell from the video how bad the pitch "nodding" was but that has always been the main cause of horizon drift in my case where flying wings are concerned. Especially affects the pitch horizon. Did it seem worse on this particular flight compared to normal ?

[b14ckyy]

You won't see the vibrations I mentioned, in the FPV fed as movement. These are vibrations, not actual plane movements. The nodding in the video is totally normal and has no effect on the AHRS at all. I flew the AR pro in very bad conditions here:
https://youtu.be/edKjR93SS7c

the only maneuver we know of that still can induce horizon drift, is a repeated fast 360° roll with short stops in between. You can see that in the video. But it recovers quickly as soon as I level out and even RTH/Loiter is still stable.

[mykls]

Hard to tell from the video how bad the pitch "nodding" was but that has always been the main cause of horizon drift in my case where flying wings are concerned. Especially affects the pitch horizon. Did it seem worse on this particular flight compared to normal ?

Pitch nodding is not that much of an issue with my AR Wing; it's more roll oscillations that I experience in bumpy conditions. Pitch nodding is an issue with my Heewing F-01 though, although it doesn't seem to have any issues with INAV 6.

Flight seemed normal up until the last minute. As b14ckyy notes, I didn't notice any vibrations in the FPV feed despite the log showing that they were there for the whole flight. I've re-built the plane now and I'll be more focused on checking for vibrations before launch now and checking stabilised mode at a safe altitude before I fly anywhere that may lead to a failsafe RTH.