Maybe a problem in the UAVCAN implementation (old and new)

ArduCopter Copter 3.6
1

Hello all the following has been tested on a PixHawk 2.1 with latest stable 3.5.7 and latest Beta Chibios 3.6 rc8

Problem:
Zubax UAVCAN GPS works as it should, barometer works too,
Magnetometer stops working as soon as i plug in any kind of RC input (SBUS and PPM tested).
This happens only when PH 2.1 is powered trough regular bec, if i power the PH 2.1 with usb cable attached to my computer everything works as it should, GPS, BAROMETER, MAGNETOMETER AND SBUS.

This thing is driving me crazy, any help or hint would be greatly appreciated.

Corrado

2

Has anybody tested CANBUS + SBUS or PPM INPUT?

Everytime i start up the system with sbus plugged in the system doesn’t initialize the CANBUS magnetometers, if i unplug the SBUS it works each time.

If i start Pixhawk with sbus unplugged and wait for the CANBUS to start and then i plug the SBUS than it works.

It looks like the RC INPUT disturbs the discovery of the CANBUS magnetometers, can any of the DEVS please take a look?

This has been tested on latest stable 3.5.7 and latest beta ChibiOS 3.6 RC 10

regards,

Corrado

3

Ok, i’ll answer myself in case it’ll be useful to someone in future.

To solve my problem i had to put a little device in line between my sbus receiver and the pixhawk (2.1). This device just waits for 15 seconds before starting to send sbus signal to rc in port. During this time canbus can initialize correctly the magnetometer on the zubax canbus device.

If i disable it, i keep getting bad campass health and canbus magnetometer is not initialized by the syste,

Maybe is my configuration or maybe is something i just can’t see at the moment but this is how i solved it.

An inside view from somebody that could enlighten me a bit would be greatly appreciated.

Corrado

4

I don’t have hardware to test this, but it seems like a strange connection. RC input in the Cube is done in the IO MCU, CAN is done in the FMU MCU. It’s certainly possible to be a software issue.

Have you tested latest master firmware? Do you have any other boards to test?

5

Hello, tested on 2 different PixHawk 2.1

Not tested latest firmware, never compiled one til now. Guess it is not that hard to do, any guide on how to? Probably time to learn something new.

Corrado

p.s. i don’t have a linux machine, can i do it from windows or maybe from ubuntu in windoows?

p.p.s. Ok found instructions on how to on WSL, will try and report back

6

You don’t need to compile it, we build it: http://firmware.ardupilot.org/Copter/latest/

7

Ah ok great, what is the name of the Pixhawk 2.1 file? fmuv3 (ChibiOS)???

Corrado

8

CubeBlack for a PH2.1 on ChibiOS.

9

You can use CubeBlack, as @james_pattison said, or fmuv3 for ChibiOS, or PX4-v3 for NuttX.

10

Tested master, same issue.

Had about 3 hours flying today on 3.5.6 RC10, canbus magnetometer gives me a bunch of variance on X and Y axis, it is fine on Z.

When i give a lot of ail or elev input the EKF turns red on mission planner.

Have to say that Zubax has very high offset numbers on calibration (in the 300 area for all axis), the other internal magnetometers have green numbers.

After the sbus bug and all this probs on canbus magnetometers i start to seriously think there is something wrong on the canbus implementation, at least on the magnetometers because both zubax GPS work flawlessy.

If you want to take a look at the driver i can do all testing necessary. I think would e a good thing to fix the problem if there is any, since a lot of new canbus mag/gps are coming out in near future (Here 2, CUAV ecc ecc)

Corrado