Drone falls while taking off

[Moutaz]

I have the NXP kit nearly a Month ago and run into issues preventing me to take off well and start flying. The problem is: after arming and trying to take off manually first by raising the throttle gradually, the drone is not stable at all and falls down in different bad ways and this is because of the motors’ speeds are not equal. I have tried every solution found on the gitbook and forums including calibrating all sensors and ESCs multiple times, check wirings and hardware connections, cleaning cables, and even flashing bootloader and PX4 again and different versions. I always check the pwm_out from QGC through mavlink inspector and the values of 4 motors are always divers and make sense that the drone cannot take off like that.

I expect at least at the start when drone is stable that the pwm signals are the same for all to have a straight take-off and then maybe depends on how tilt the drone is, the pwm_outs will be a bit different as a try to balance the drone again in the air. Any ideas what could be wrong here? 

[GeraldP]

Hello @Moutaz , @Shake21,
Please share you logfile of PX4 to be able to help. @Shake21 with just a video from this distance it's hard to judge anything.
But to me it seems that too less throttle input is given. You have to understand that ground effect (circular airflow near ground) could influence your vehicle heavily. When i'm taking off i always give a fast throttle punch of >75% trottle and reducing that quickly to 50% (assuming your hover throttle is correct) when the drone is about 1m above ground.

Other topics to check always

• is the motor spin direction correct? Are the propellers mounted correct (direction & tightened)? front left - CW , front right - CCW, back right - CW, back left - CCW
• is the motor/ESC assignment correct? https://nxp.gitbook.io/hovergames/userguide/assembly/connecting-all-fmu-wires#servorail
• has the sensor calibration being done properly? https://nxp.gitbook.io/hovergames/userguide/qgroundcontrol/sensors
  You both stated yes, checked.
• Has the radio controller setup being done properly? https://nxp.gitbook.io/hovergames/userguide/qgroundcontrol/radio-and-flight-modes
• Has the ESC calibration being done properly? https://docs.px4.io/main/en/advanced_config/esc_calibration.html
  You both stated yes, but please check the procedure carefully you need to wait until the ESC have beeped multple times. Cant recall the exact beeping sequence at end of calibration.
• Is takeoff beeing done in manual flight mode?

If all the above have been checked and ok the only reasoning i see is the throttle input. Further help can only be done with a log file - https://docs.px4.io/main/en/getting_started/flight_reporting.html#sharing-the-log-files-for-review-b...

And a final word on safety: if you are an inexperienced drone pilot never do you first tests indoors and or close to people. Better try outdoors at an open field with nobody else around.
Testing on a soft gras also reduces the likelyhood of a damaged propeller.

[Moutaz]

Hi @GeraldP ,

Here is the log file of one of the tries (I also tried with hovergames airframe if you suspect the frame type) ==> https://logs.px4.io/plot_app?log=379457e4-d2e6-4118-8b22-3574d0fbfce6

Could you please answer me about the specific pwm point? in qgc MAVLink inspector, I observe the values of servo1,2,3,4_raw in the tab "SERVO_OUTPUT_RAW" to see the pwm values decided by the flight controller which can change according to the drone tilting/direction in order to rebalance and make the drone stable. On attempting to take off from a level ground, I see the PWM of 4 motors are not close to each other and I suspect that is the reason for the instability (I expect them to be the same or at least close to have similar power to all motors and makes a stable take off). What do you think about that? Because of that I suspected the fmu itself as the sensors seemed weird to me or sth but I get the same behavior with a new fmu..

I hope you can help me on that point as I am really desperate to make it work for my thesis..

[Shake21]

Hello @Moutaz ,

Greetings, I apparently am also facing the same problem. Did you overcome this problem? If yes how did you manage to do it? Really looking for your response.

I really appreciate any help you can provide.

[Moutaz]

Hi, unfortunately I have not yet been able to solve the problem! I replaced ESCs, motors and the FMU itself but it still does not fucntion properly. I have exactly the same behavior in your video although recently it is a bit better (lift off the ground unstably for 2-3 seconds and then falls off.

I believe the problem is about the FMU itself as the PWM values sent from the controller do not make much sense for me and it causes unequal motors' speed. (see MAVLink inspector tab in QGC and then tap "servo_output_raw".

[Shake21]

Hi, @Moutaz,

Firstly thanks to a few of my friends, who helped me make the drone fly. Secondly, I know you must have heard everywhere about the recalibrations of motors, ESC's, compasses, and accelerometers. If everything is calibrated then, check the wires from ESC to FMU (the black and white ones) are in the correct order. Also, it's really not necessary for the PWM output to be the same for all motors. And make sure when you calibrate the drone you do it outside (where it is safe to take off). Also, I have attached a picture that might help you in understanding the motor and propeller direction. The wire connection(black and white) from the ES to FMU should be in order according to the picture given below. 

https://docs.px4.io/v1.9.0/en/flying/basic_flying.html

I have attached px4 doc link that also really helped me, please go through it. 

I hope I have explained this sufficiently. And Good luck with your project.

[Moutaz]

Hi @Shake21 , 

I have tried all what you mentioned but still .. I still believe the PWM need to be similar, okay not exactly but to be similar to make sense. Because it behaves for me much better in Altitude mode where the pwm values for all go high at once with all similar values around 1800 and so. But still moves slowly to a specific direction without input from me. So, I make it land quickly before it hits the wall or sth.

Could you please help me in last thing and tell me your pwm values in QGC MAVLink inspector and then tap "servo_output_raw" when you arm the vehicle and start increasing the throttle up ? in manual or altitude mode.

[Andrew_M]

If you seeing it slowly drifts instead of hovering in place, can you tell me if it is always the same direction (forward, backward, etc) and does it behave the same if you rotate 90 degrees? (Face it North instead of East, etc)

[Moutaz]

Yes always the same direction because again: It is all about the weird PWM values coming from the controller .., direction depends on which motors have more power

[GeraldP]

The "weird PWM" is just a result of what is happening within the SW.
When looking at your log Flight Review - Quadrotor (px4.io) i see three things:
1) Pitch angle is at -2 deg
2) Pitch angular rate setpoint ~ 15deg/s
3) Pitch actuator control is +0.04

In other words. The drone thinks it's pitched backwards and tries to compensate by pitching forward. To me it looks like that the level calabration has not been done properly.
If you arm your drone and do not take off immediately by a high trottle input a fall over can happen in such a situation. Also moving the drone after plugging the battery in needs to be avoided. At bootup some calibration is happening that can lead to such a behaviour.

Other thing i notices is that you have very high vibrationes. Is your FMU mounted on the dampening plate?
And it seems you are not using a stable viersion of PX4. For a first test of the drone please always use one of the PX4 stable releases!
For further analyseis pictures of your drone, especially the FMU mounting and cabeling around is needed. As well as a video of a take-off attempt with the related log-file.

Are u using the RC control? cause i don't see any RC input in the Log!

[Moutaz]

@GeraldP 

Hi, Please ignore the previous log file as it was not a real attempt. Here is the recent attempt with everything (cabeling, calibration, pitch angle, etc..) taken care of and considered.

https://logs.px4.io/plot_app?log=b1877d16-243b-4985-b255-5af19768caa7

Please have a look at this here and tell me why it could behave like that? Find the attempt video and some pics of the setup attached. I would appreciate your help so much!

[GeraldP]

Sorry to say, but there is nothing wrong with the behaviour of your drone. It's just the consequence of trying it indoors just a few centimeter above ground. As i have stated previously there is the issue of ground effect if you are hovering just some cm above ground. Your drone will ALWAYS move uncontrolled in different directions. You REALLY have to fly ~1m above ground to get out of that!

On the "high" PWM outs. A drone controls it's position by doing FAST changes of the thrust. So constantly the RPM is changing of every motor. This is NORMAL.

What i see at your drone to improve:
1) please ALWAYS mount the GPS. It has an additional magnetometer on board that is less disturbed by magnetic field of ESC than the FMU onboard magnetometer. Even for flying indoors. Please use the shorter rod of the GPS mounting.
2) please use manual/stab flightmode instead of Altitude.
3) tuning of your drone has not been done yet. It's a bit wobbly still.
4) your vibrations are still too hight. That could result in being wobbly too.
5) one of the rootcauses of the high vibrations could be the long, not secured cable, of the telemetry radio. Cables should be ALWAYS running straigt downwards from the connector to the frame and secured via zip-tie to have the lenght of a loose cable as short as possible.
At a drone everything needs to be tightened to reduce vibrations. Cables, Landing gear,....

[Moutaz]

Hi @GeraldP ,

Thank you for these points, I will try them asap. The point about tuning is a bit confusing for me because when I read about it, it said the pre-condition for such tuning is that the drone is at least flying(hovering) stably but this is not the case for me so I guess tuning now is hard to do, or am I missing something here?

[GeraldP]

Hi Moutaz, i think you are expecting too much. To me your drone flies stable under the current circumstances. I have done tuning on much wobblier systems in the past.
First of all you should get the hover throttle right. Again, please try outdoors if possible or in a bigger indoor location. I know outdoors is not possible for you being that close to the airport in Finkenwerder.
I can really recomment looking for a RC model airfield and asking for permission to test there.
And please be aware of the rules when flying outdoors -  Drohnen Gesetz: Vorschriften / Genehmigungen / Verbote as well as Luftfahrt Bundesamt - Unbemannte Luftfahrtsysteme (lba.de)

[Shake21]

@GeraldP , @_Leo_ , and @ all the NXP EMPLOYEES

Please reply to the questionnaires, It is very disappointing that customer support is not at all bothered to reply to our doubts and problem. There is no use of discussion forms if the support teams do not help or respond at all.