PX4-Autopilot: Holybro ESC problems (can't fly with master)

Unfortunately I can’t continue the bisect with the s500 until I get some new props. I’m going to setup my 250 airframe in the mean time and see if I can’t reproduce the issue. Travelling this week so I’ll be back on it the following week.

On Sun, Nov 3, 2019, 1:49 PM Mathieu Bresciani notifications@github.com wrote:

@mrpollo https://github.com/mrpollo I think we still need to find the cause of the issue, lowering the cutoff just hides the underlying possible issue. This is why the bisect is really important to helping us finding the modifications that leads to the change of behavior.

@dakejahl https://github.com/dakejahl any news?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/PX4/Firmware/issues/13147?email_source=notifications&email_token=AI27PPTLESLK6V75C5J62K3QR42MDA5CNFSM4I7DKCI2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEC54MZQ#issuecomment-549176934, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI27PPRADRY6GPYL2EVJWUDQR42MDANCNFSM4I7DKCIQ .

That’s not really concerning because we’ve changed the filtering pipeline to accommodate high-performance racers so that the cutoff setting is more effective. I’ll let @bkueng / @bresch chime in, but that frame has a lot of flex and so what I’m seeing here is what I would expect with 30 vs 80 Hz cutoff.

Changing IMU_GYRO_CUTOFF from 80 back to 30 solved the issue 😕

I can’t remember why I changed it away from default in the first place… 🤷‍♂️ (I think it was early on during bringup and it was flying like crap and someone suggested to change the gyro cutoff). But still concerning that there is such a large difference in behavior between two versions. Any ideas?

https://review.px4.io/plot_app?log=b214a366-11a8-4ca3-9850-0faa3fd62c57

Here’s a bench test with no props attached, I arm the vehicle and let it idle for a little bit, then ramp up the throttle to 0.5 and then back down to idle.

1.9.2 https://review.px4.io/plot_app?log=bf9f4248-a7e5-42dd-9f83-e60df970ad0a

master https://review.px4.io/plot_app?log=eccaa7ec-6fb8-4e93-88dd-9689243c1c05

You can clearly see the raw gyro/accel go absolutely nuts on master as soon as there is more than idle throttle.

Now the hard part here is determining cause/effect. I did a diff git diff v1.9.2 master -- src/drivers/imu/mpu6000/ and nothing changes functionally with the driver (20689 is crammed in with mpu600). I verified on the bench that the sensor is being selected correctly and that the data more or less looks the same when you poke it. Also notice the vibration levels between 1.9.2 and master are pretty much the same at idle throttle, the data really only gets bad when you throttle up.

This leads me to believe the crazy vibrations are real vibrations – caused by the shuddering of the motors (Video of shuddering). If you take a closer look at the actuator outputs, you can see on master the output values do not increase smoothly, but are rather jagged.

1.9.2 master

Yes I will investigate this early next week when I am back in town.

@dakejahl mentioned he has the problem on fmu

Sorry I think I misunderstood the question on the dev-call, I am indeed connected via IO PWM OUT

I have requested hardware so that I can help with this. but I will not be able to work on it till after I return on the 16th.