Crash Log File Assistance

ArduCopter
1

Hello you log file geniuses. I’m not well versed in analyzing logs yet and could use some assistance in trying to figure out what caused a fairly gnarly crash this morning.

I started the tuning process and about 2 minutes into the flight, the copter started continuously yawing clockwise pretty fast and I had no choice but to put her in the dirt. The arms and motor mounts were tight as far as I could tell so I don’t think that was the issue. One of the motor arms is now twisted quite a bit but that seems likely from where the majority of ground impact happened (it’s also snapped in half.)

The uncontrolled yaw did happen yesterday briefly in RTH and was able to recover in stabilized or loiter mode (can’t remember which) and the copter flew just fine after that. Today was a completely different story.

https://drive.google.com/file/d/1qePmURZmhzbtFjutFHr-fsRrm_5rnkPZ/view?usp=sharing

2

I think to start with there was a bit of a yaw imbalance already, meaning there is likely a twist in the arms. However for the crash I’m going to bet one of the arms came loose or rotated because the CCW pair of the motors all but stopped in flight but the CW pair kept going. It’s probably not a motor/esc failure because that is usually marked by the bad arm going to full power and the opposite arm going to min. I don’t see that here.

Servos_1-4
Servos_1-41200×600 124 KB

3

Ok interesting thank you! The arm that looks like it may have been the culprit is CW #3 as it’s twisted about 45 degrees CCW at the frame which would account for the spin. I wonder if the brief moment yesterday possibly corrected itself and re-presented today. It’s odd though because this arm was definitely tight. The mounts I am using have a 6 degree up tilt which probably added to the problem. The frame and motor centers of inertia are calculated to give a true quad layout. Where else can we look in the log for other possible or additional causes?

4

I would be worried about all those “fast fifo reset 3” messages. You have 2 Dynamic Notch filters enabled with the same settings for a Dynamic FFT referenced notch. Disable the 2nd one. Perhaps the load is causing the reset error.
I’m not suggesting this is the cause of the crash but you don’t want these types of errors.
And install 4.3.3 you are still on the rc1 release.

5

Ok yes I had enabled FFT to start with tuning today. That may have been the ultimate reason for the crash but it was not enabled yesterday when the same spin happened that I recovered from. I’ll take a look at the messages. Thank you.

6

No problem with that but disable the 2nd Notch. set INS_HNTCH2 to 0.

The fifo reset is actually a fix put in place for the ICM20602 IMU. W/o that bad things could happen. So, it’s doing what it’s supposed to do. If you flash 4.3.3 stable there will be less of these messages generated.

7

I’ve made all the changes suggested and while I’m waiting for parts to come in, is there anything else I should explore in the log to rule out other possible causes for the crash? I’m fairly certain a twisted arm was the cause but I’d like to double check everything.

8

What motors, props and ESCs do you have? Is it the integrated T-Motor mount/motor/ESC ?
What battery? 6S LiPo?

This is what I would do:
ATC_ANG_PIT_P,6.0
ATC_ANG_RLL_P,6.0
ATC_INPUT_TC,0.20
ATC_RAT_PIT_I,0.10
ATC_RAT_PIT_P,0.10
ATC_RAT_RLL_I,0.10
ATC_RAT_RLL_P,0.10
ATC_RAT_YAW_P,0.08
ATC_RAT_YAW_P,0.8
ATC_THR_MIX_MAN,0.5
BATT_ARM_VOLT,22.10
BATT_CRT_VOLT,21.00
BATT_FS_CRT_ACT,1
BATT_LOW_VOLT,21.60
BRD_BOOT_DELAY,3000
FFT_ENABLE,0
GPS_GNSS_MODE,7
GPS_GNSS_MODE2,7
INS_HNTC2_ENABLE,0
INS_HNTCH_BW,20
INS_HNTCH_FM_RAT,0.7
INS_HNTCH_FREQ,40
INS_HNTCH_MODE,1
INS_HNTCH_REF,0.12
INS_LOG_BAT_MASK,1
INS_LOG_BAT_OPT,4
LOG_BITMASK,180222 ← PIDs
MOT_BAT_VOLT_MAX,25.20
MOT_BAT_VOLT_MIN,19.80
MOT_THST_EXPO,0.4 ← depends on ESC/motor comination !
PILOT_THR_BHV,7 ← suits spring-centered throttle

Copy/paste this into Notepad
Strip out the comment bits
Save as {somethingname}.param
In MissionPlanner, Full parameters , Load from file

9

Hi Shawn, I am running T-motor integrated 6007’s with the Alpha’s for now on 6S. 22.2 props.

10

OK, use all the parameters I provided

11

Great thank you! I’ll start working with those once I’ve rebuilt.

12

ATC_RAT_YAW_P, 0.08 I’m assuming? I see that param listed twice at 0.08 and 0.8

13

Yes, sorry, it should be
ATC_RAT_YAW_I,0.08
ATC_RAT_YAW_P,0.8

1 Like
14

@xfacta I completed a test flight with the params you recommended. While I didn’t want to risk anything after my previous crash I noticed a few things. First was that the learned hover value was much higher than previous. There were oscillations in yaw, everything felt twitchy, and altitude wasn’t stable. It would lose and gain altitude. Not sure if that’s what people refer to as toilet bowling.

I’m going to review the log now so I learn but would appreciate some input.

https://drive.google.com/file/d/1-g_ffIFQHntqFjj52AOGvtr1cqBK5EgX/view?usp=sharing

15

@tridge I’m a bit concerned about all the IMU0 fast fifo reset messages in this log. There is 43 of them. I can see there’s a PR to limit the rate of messages, but I’m not sure what’s causing them.
Cube Orange with 4.3.3

Toilet-bowling is a circular motion usually associated with high X and Y axis vibrations and sometimes a wandering GPS position.

From that log, adjust these
ATC_RAT_PIT_D,0.0025
ATC_RAT_RLL_D 0.0025
BATT_ARM_VOLT,22.10 ← you have it lower than LOW_VOLT
INS_ACCEL_FILTER,10
INS_HNTCH_FREQ,46
INS_HNTCH_BW,23
INS_HNTCH_HMNCS,7
INS_HNTCH_REF,0.17

These will settle down the position controller a bit, stop chasing the GPS position so hard. They are safe and can easily be put back to defaults if you need more accurate position holding in the future. They are just half of original values.
PSC_POSXY_P,0.5
PSC_VELXY_D,0.25
PSC_VELXY_I,0.5
PSC_VELXY_P,1.0

There is a slight physical yaw imbalance, like a motor mount is slightly twisted. Generally motors 1 and 2 (CCW) have to work a bit harder than motors 3 and 4 (CW). See if you can perfectly align all the motors/props so they are spinning on exactly the same plane.

image
image1658×837 20.9 KB

Set everything in preparation for the next test flight, but hold off on flying until we’ve got more information on the “IMU0 fast fifo reset” issue.

If the next test flight is OK or better than before, I would tentatively try these:
ATC_RAT_PIT_I,0.09
ATC_RAT_PIT_P,0.10
ATC_RAT_RLL_I,0.09
ATC_RAT_RLL_P,0.11
Pitch and roll are behaving slightly differently.
If there is reasonable stability and response to RC inputs then Autotune can probably sort out the details.

16

Ok thank you Shawn. I have the new values set and I missed the batt_arm_volt first time around. I’ll align the motors again. I think #3 may be off slightly while the others are within .2 of a degree of center. I’ll wait for the go-ahead regarding the fifo reset issue. Think there’s a chance the Cube Orange could be defective?

I should mention that the arms do have a 6º up tilt.

17

I also meant to add, you dont seem to have those issues, at least not bad.
Once there is good stability and flight can be performed confidently, do a flight with plenty of yaw and circular patterns and we can run that log through MagFit to improve the compass settings too.