Need help with Toilet Bowl issue on 5" props quadcopter

ArduCopter Copter 4.0
15

@xfacta @Eosbandi sorry for the delay, was involved with some other tasks

So I have autotuned my 5" quadcopter running arducopter 4.0.4, these are params i set before autotuning:
ACRO_YAW_P,3.5
ATC_THR_MIX_MAN,0.1
ATC_ACCEL_P_MAX,165300
ATC_ACCEL_R_MAX,165300
ATC_ACCEL_Y_MAX,31500
ATC_RAT_PIT_FLTD,37.5
ATC_RAT_PIT_FLTT,37.5
ATC_RAT_PIT_FLTE,0
ATC_RAT_RLL_FLTD,37.5
ATC_RAT_RLL_FLTT,37.5
ATC_RAT_RLL_FLTE,0
ATC_RAT_YAW_FLTD,0
ATC_RAT_YAW_FLTT,37.5
ATC_RAT_YAW_FLTE,2
MOT_THST_EXPO,0.55
MOT_THST_HOVER,0.2
INS_ACCEL_FILTER,20
INS_GYRO_FILTER,75
MOT_BAT_VOLT_MAX,8.4
MOT_BAT_VOLT_MIN,6.6

After autotuning in all roll,pitch and yaw axes these are some params:
ATC_ANG_PIT_P,18
ATC_ANG_RLL_P,18
ATC_ANG_YAW_P,6.686522
ATC_RAT_PIT_D,0.008305321
ATC_RAT_RLL_D,0.007172775
AUTOTUNE_MIN_D,0.001

These are the PIDs after autotuning:

Screenshot from 2020-10-05 11-55-57
Screenshot from 2020-10-05 11-55-571366×768 167 KB

After looking at autotuning guide, these values suggest that autotune has been successfully completed, what you think about the tuning on this quadcopter ?

Log and complete parameter list is attached below, this is the log of flight after finishing autotuning:
5" quad log

5" quad parameters

16

That must be a very good tune now - the ANG P values went to maximum, you could even lower those a bit if you like without too much disadvantage.

There still a little bit of movement in flight, but it’s only 0.2 of a degree and probably less that nearly every other aircraft:

image
image912×354 35.8 KB

The rest of the time pitch and roll are tracking Desired very well.

I would highly recommend doing the harmonic notch filter tuning if you can, then run another Autotune.

1 Like
17

Gotcha, will see if i can do harmonic notch filter tuning on 1MB flash boards, thanks for helping out

18

@xfacta
When running autotune multiple times on same 5" props quadcopter then after autotuning in roll and pitch, PI values for both of them have tendency to vary from 0.15 to 0.25, and i am also getting quite a variance too between their behaviours, sometimes both PI values for Roll and pitch are quite close but sometimes roll PI comes out to be 0.15 but at same time pitch PI comes out to be 0.25, and i autotune when weather is good and winds are calm, still there is quite a variance between roll and pitch PI values when autotuning multiple times, is it okay to have this much variance between PI values in roll & pitch ?

19

@Leonardthall always says that a 25% variation is normal. One thing to watch out for is that the tune on one axis will affect the tune on other axis, so the order you do things in will make a difference.

20

I think, you can solve this problem via In-Flight Compass Calibration.

  1. Set COMPASS_LEARN parameter to 3.
  2. Arm and flight in AltHold mode.
  3. Fly until you see “CompassLearn: finished” messages on your GCS.
  4. Land and Restart your copter.
21

Yeah, i know about this feature, but i generally calibrate compass in open field, i doubt how much it will help in dealing with toilet bowl effect, i thought in-flight compass learning was mainly for bigger drones as it is difficult to calibrate them manually

22

In this page at Overview part(Advanced Compass Setup — Copter documentation) you can see this description.

Without an accurate heading the vehicle will not move in the correct direction in autopilot modes (i.e. AUTO, LOITER, PosHold, RTL, etc). This can lead to circling (aka “toiletbowling”) or fly-aways."

So, the vehicle must have accurate heading to ignore toilet bowl effect.

Since the magnetic field will be different when the motors are running, in my opinion, in flight calibration will solve your problem. Can you make a test?

23

Definitely, i actually just did compass/motor calibration as well, it took some time in delivery of power analyzer, that’s why i haven’t updated people on this thread yet, i do have some inputs incoming soon, but now you have also suggested in-flight compass calibration, so i think i should not let any stone unturned, i will try that tomorrow, will see if i can perform both compass/motor calibration and in-flight compass learning on multiple 5" quadcopters then i will have more data available to analyse this toilet bowling issue, @xfacta will probably take a day or 2 more for logs, not able to do harmonic notch filter tuning in upcoming logs but it is definitely in my bucket list

@andyp1per But overall i can also see on my small 5" quadcopters that total field strength is around or more than 500 mGauss, sometimes it even cross the limit of 550 mGauss, which suggest that i probably need to work with my setup too, i am not sure about that, i am using external mag of RTK GPS but still overall field strength looks pretty large

24

On a small quad there will be a lot of interference, you defintiely want to do COMPASSMOT

25

i did that today first time, but interference comes out to be max 4% at around 70-75% throttle using current sensor, i just did that on one quadcopter for now, had around 14-15 flights after that, found just slight curvy movements in only one particular flight, but rest of the flights seems okay, i have the logs of todays flights, i can share them if you wanna have a look, cause i doubt if compass/motor calibration went well or not, i was really expecting more than 15-20% interference, anyway i am sharing logs here in few minutes

26

@andyp1per @xfacta @Eosbandi All these logs are of same 5" quadcopter with 3 different sets of similar 2 cell 18500 mAh Li-Ion batteries, have implemented compass/motor calibration using current sensor

FIRST LOG

COMPASS_AUTO_ROT = 2
COMPASS_AUTODEC = 1
EK3_MAG_CAL = 3
https://drive.google.com/file/d/1gvbKjCjwX6qqVospvJtgEwJOLeSpHNNy/view?usp=sharing

SECOND LOG

COMPASS_AUTO_ROT = 0
COMPASS_AUTODEC = 0
EK3_MAG_CAL = 2
https://drive.google.com/file/d/1c5K0V0iWVtfsyML6qd0NO9UwaJlRy30s/view?usp=sharing

THIRD LOG

COMPASS_AUTO_ROT = 0
COMPASS_AUTODEC = 0
EK3_MAG_CAL = 0
https://drive.google.com/file/d/1CBJXzlpLM6OZNxnbpGBZPzrohAwXtRHl/view?usp=sharing

These parameters change does not seems to have any other effect other than stopping default behaviour of ekf yaw reset after takeoffs, that’s why i was looking to know what’s the difference between heading fusion and 3-axis fusion, as i cannot able to find any difference from the logs and flights

In all these 3 logs, total magnetic field strength was around or more than 550 mGauss, and i don’t know how acceptable is the variance of field strength in these logs

27

Are you calibrating with props on?

28

yup, had switched the props in order to make quadcopter pushing down to the ground when raising the throttle

29

@AGurel things got little worse after in-flight compass learning with compass calibration fitness set to 8 (strict), 2 of the 5 small 5" Quadcopters continuosly doing toilet bowling in most flights, though offsets didn’t varied much compared to manual calibration, there was around ±(5-30) change in the values of offsets

Have done CompassMot calibration using current sensor, though there was just 4-7% of interference detected by this calibration in multiple 5" quads, done in-flight compass calibration even in strict fitness still this toilet bowl issue just getting more ramp up than before

I tried everything which is there for managing magnetic interference on compass, but still not able to manage this issue, only option i can think right now is to strictly review and made changes in my quadcopter ESCs and wiring setup

30

Can you post a picture of your build.

31

Will definitely share the detail images by tomorrow

32

Top view and rest all 4 side view build pictures are here in this link:
https://drive.google.com/drive/folders/1q2ihSHhd9CdObJSN9LnRzf-XQv9Ab1hd?usp=sharing

Hardware Details:

  • Airbot Omnibus F4 SD (V1) running ArduCopter v4.0.4
  • T motors F20
  • 4 in 1 ESC 15 Amps
  • PPM receiver
  • Ebyte E62
  • Custom build M8P RTK GPS

In our custom build M8P RTK GPS, 2 connectors are there for separate connection to mag and gps, rest everything is similar to HERE+ RTK GPS, position of mag is just below the GPS PCB under forward facing marker on GPS. I don’t think our custom build GPS is an issue cause we using same GPSes on different drones without issues. Also, when replacing our custom GPS with HERE+ GPS on drones which causing toilet bowling even then toilet bowling issue still persist on those drones.

Overall, my all small 5" quads are quite prone to this toilet bowl issue even after doing compassmot calibration and in-flight compass learning in strict fitness.

It will be really helpful to get some more inputs about our quad build cause compassmot calibration not showing much interference from motors (about 4-7%) and in-flight compass learning also not changing the offsets too much compared to manual compass calibration. That’s making me think now that quadcopter build might be the major issue for this toilet bowling.

33

There’s no way you should be getting much MAG interference with it mounted up there, my guess is something else. How hot does the flight controller get?

34

never checked the temperature, i will try to meausure when fly this same quad today, but i also doubt about the onboard voltage as there is no board voltage sensor on this board, i doubt if this board getting sufficient voltage or not, can check this only on ground, will update about it too, but i don’t know if this can also be one of the reason for toilet bowling or not, also overall magnetic field strength on my 5" quad are quite high around or more than 500-550 mGauss