Motors stuttering and turning off at high trusts

[Thomas de Rooij]

Hi all!

This is my first post so I hope I'm providing enough information. I'm a student working with DroneTeam Twente to build a drone to join the UAS challenge but we've run into a problem. Our drone is a Funcub model, but we're testing it in quad mode. During our first test flight we attempted to take off but at high throttle, about 80%, some motors (differs per test which ones) start to stutter and then turn off. Watch

Our setup is as follows:
- PX4 FC
- DYS Aria 70A Blheli_32 ESCs
- HolyBro PM07 V2.3 power management board
- QX motor QA 2825 700KV motors
- 2 6S batteries in parallel

Our ESC signal and ground go straight into the PX4, since earlier we had issues with a ground loop when connecting the ground of the power supply of the ESC to the ground of the signal of the ESC. This lead to heavy stuttering, which seemed to be solved until now. The actual problem: In the attached video (click here to watch it) you can see one motor has a prop (it is turned around so it pushes into the ground for testing purposes) and the others don't, the ones that don't work fine.
Our ESC settings are as follows: .
We've tried changing the Rampup power, Demag compensation & motor timing, to . Our ESCs are rated for 70A and can handle 6S batteries, so this shouldn't be a problem.
Some more things to note: when activating the motors in QGroundControl this still happens.
Any clues how we could fix this or measure where it's going wrong? Any help is appreciated!

 

[LoneRCRanger]

Sorry, your video link is not present. When you want a link to appear, click the chain link, enter the URL and a pseudo link to have the URL assigned to. Viewer would then click that link to view your video.

 

[wafflejock]

Curious what protocol is being used for the pixhawk FC to escs on the signal wire (PWM, or DShot600 or something else?). Believe pwm is more susceptible to noise and requires calibration (look up esc calibration if curious or haven't tried that yet). I can see the video link now too btw. Does this only happen when the motor is "loaded" with a prop on.

 

[wafflejock]

Ultimately it does seem like an issue with the esc(s) maybe try flash them back to defaults and test unloaded (no props) see if the extra current draw from prop is making any difference. Otherwise would try another esc really, you can try diagnosing the issue with an oscilloscope but will be faster/cheaper to just start replacing parts (guess and check).

 

[wafflejock]

If prop on effects things it could either be that extra load on the esc is too much or that the prop is imbalanced, after running it can also check if the motor is warm or hot, warm is okay hot is bad, if you can barely touch it then the motor is getting more current put through it then it can dissipate which will mess with lubricant in bearings etc.

 

[Thomas de Rooij]

The video link is updated and should now work! As you can see, the problem only occurs when the prop is loaded.
We've tried ESC calibration and flashed them too. When hooking up the EScs and motors to a servo tester they work fine, so we're suspecting it has something to do with the PWM signal that the PX4 is emitting. (So to answer your question, we're using a PWM signal).
We also routed the ground and signal of the ESC outside of the drone to the PX4 to prevent noise from other cables.
We also considered it might be the temperatures of the ESC or motor, but these are fine, so we're assuming the problem is with the PX4 emittting the PWM signal.
We're going to check what the signal looks like and update the firmware and replace the PX4 to see if that fixes anything.

 

[Thomas de Rooij]

For those still interested: we've found the problem. Using an oscilloscope we measured the PWM signal coming out of the PX4, which was fine and looked clean. Then we measured the PWM signal near our ESC and this had a lot of noise when the prop was attached and on 80% throttle. We did some digging and when we ground the signal ground to a different battery then the one powering the PX4, the noise disappears.
We drew the conclusion that we are dealing with a ground loop (again...) and are now looking at ways to either filter the noise or eliminate the ground loop. Thanks for the responses on this thread!