Holybro-KAKUTEH7-V1's magnetic sensor problem

[kevinuav]

Describe the bug

I find the kakuteh7's magnetic sensor may have issue. My kakuteh7-v1 is installed in my fixed-wings. I connect an external magnetic sensor for it since the FC has no MAG inside. However. The estimated direction is almost reverse to the real. I have tried to correct it with the magnetic sensor setting. But it doesn't work. It should be a bug in the transplantation of the kakuteh7.

To Reproduce

When you connect the i2c external magnetic sensor for kakuteh7-v1. You will find its estimated direction is almost 160 degrees reverse to the real. And it can't be corrected with the magnetic installation parameters.

Expected behavior

The estimated direction of the magnetic compass is almost 160 degrees reverses to the real when the vehicle keeps still (the GPS direction is invalid). This is critical for the plane's automatic takeoff.

Screenshot / Media

No response

Flight Log

At about 2:15 to 2:18. You can see the magnetic estimated direction suddenly changes to the real direction when my plane is taking off.(The GPS direction is valid).

https://logs.px4.io/plot_app?log=8405692c-0f61-46ee-834c-41338c26e5eb

This is the data comes from the other plane of kakutah7-v1. It's estimated direction is almost 80 degrees different from the real during the taking off procedure at about 3:15-3:40. Until the GPS's direction becomes valid.

https://logs.px4.io/plot_app?log=c8350813-9f05-4eb7-8937-c60f849dbe14

Software Version

V1.14.0-main

Flight controller

Holybro-kakuteh7-v1

Vehicle type

Fixed Wing

How are the different components wired up (including port information)

No response

Additional context

No response

[bresch]

The magnetometer data looks weird: it's almost a constant value with noise (https://logs.px4.io/plot_app?log=8405692c-0f61-46ee-834c-41338c26e5eb#Nav-Raw-Magnetic-Field-Strength). Are you sure it's working?

[kevinuav]

The magnetometer data looks weird: it's almost a constant value with noise (https://logs.px4.io/plot_app?log=8405692c-0f61-46ee-834c-41338c26e5eb#Nav-Raw-Magnetic-Field-Strength). Are you sure it's working?

It can be calibrated successful. The FC can read the data from it.

[julianoes]

@bresch us right, something is wrong with the data. I'll check with my board

[julianoes]

So what I missed is that the KakuteH7 v1 does not have a mag included.

@kevinuav can you post a picture of the mag that you're using? I need to check if I have one that is the same.

[kevinuav]

This is the data comes from the other plane of kakutah7-v1. It's estimated direction is almost 80 degrees different from the real during the taking off procedure at about 3:15-3:40. Until the GPS's direction becomes valid.

https://logs.px4.io/plot_app?log=c8350813-9f05-4eb7-8937-c60f849dbe14

[kevinuav]

So what I missed is that the KakuteH7 v1 does not have a mag included.

@kevinuav can you post a picture of the mag that you're using? I need to check if I have one that is the same.

OK, I will take a photo for it. It's Holybro Nano Ublox M8 5883GPS Module. Which has an i2c MAG sensor inside.

[julianoes]

I don't think I have that one.

But it looks like it has the qmc5883l sensor.

Can you try qmc5883l status in MAVLink shell?

[kevinuav]

I don't think I have that one.

But it looks like it has the qmc5883l sensor.

Can you try qmc5883l status in MAVLink shell?

nsh> qmc5883l status INFO [SPI_I2C] Running on I2C Bus 1, Address 0x0D qmc5883l: reset: 0 events qmc5883l: bad register: 0 events qmc5883l: bad transfer: 0 events nsh> qmc5883l status

[julianoes]

Nothing bad showing there. That's odd, and I'm afraid I can't reproduce it here because I don't have the hardware.

[bresch]

This is the data comes from the other plane of kakutah7-v1. It's estimated direction is almost 80 degrees different from the real during the taking off procedure at about 3:15-3:40. Until the GPS's direction becomes valid. https://logs.px4.io/plot_app?log=c8350813-9f05-4eb7-8937-c60f849dbe14

In this log, the norm of the magnetic field is far from being constant (it should be a straight line with a bit of noise). Looking at the calibration parameters, the Y offset indicates that something produces a quite strong magnetic field near the magnetometer: CAL_MAG0_YOFF, 1.576

[kevinuav]

Hi. I got a log which clearly shows the issue of the kakuteh7's magnetic sensor I mentioned before. The plane "side slips" completely after take off(5:39). Then it suddenly corrected at 5:58 itself. Here is the log: https://review.px4.io/plot_app?log=730fe5ec-7059-477f-98b2-f96cc8ec942f