search

kriswiner / MPU9250

Arduino sketches for MPU9250 9DoF with AHRS sensor fusion
1.04k stars 471 forks source link

MPU9250 reproducable yaw error #298

Open jhergeth opened 6 years ago

jhergeth commented 6 years ago

Hello, as I am using the MPU9250 with an ESP8266 I had to adapt code from different sources. I am using the code from Kris to calculate the quaterions, an enhanced version of the Sparkfun-Library and the wikipedia code to calc the euler angles. I get data from the chip at 100 Hz (polling) and calculate the MahonyQuaternion every 3 ms. I also included the mag sensor calibration from Kris. And I use the correct declination for my place.

The resulting euler angles are stable and reproducable. Rotating the device forth and back gives results in the same angles. So I am quite happy with that! Thanks for all your postings Kris, they helped a lot.

But............ when I compare the absolute values to the real values (I know how much I turned the device), I get a reprducable error. It seems not to be influnced by calibration and looks like this: grafik

Around 0° it is roughly 10%, but increases towards 180°.

Any suggestions? The result is the same when I use the MadgwickQuaternion calc. In the Mahony/Madgwick routines are some free parameters Ki and Kp. Should I change them? Or GyroMeasError/~Drift?

Thanks for your work! Joachim

kriswiner commented 6 years ago

Not sure what the trouble is. Quats every 3 ms means a fusion rate of ~300 Hz? Not very fast. I think you could benefit from increasing the sensor sample rate to 400 Hz or even 1 kHz. But the fusion rate is really to slow. Shoud be 10x the sample rate or so.

I usually get 2 kHz fusion rate with the ESP8285, see:

https://github.com/kriswiner/ESP8285/tree/master/MPU9250

The fusion parameters don't matter...

On Sun, Jul 22, 2018 at 9:06 AM, Joachim Hergeth notifications@github.com wrote:

Hello, as I am using the MPU9250 with an ESP8266 I had to adapt code from different sources. I am using the code from Kris to calculate the quaterions, an enhanced version of the Sparkfun-Library and the wikipedia code to calc the euler angles. I get data from the chip at 100 Hz (polling) and calculate the MahonyQuaternion every 3 ms. I also included the mag sensor calibration from Kris. And I use the correct declination for my place.

The resulting euler angles are stable and reproducable. Rotating the device forth and back gives results in the same angles. So I am quite happy with that! Thanks for all your postings Kris, they helped a lot.

But............ when I compare the absolute values to the real values (I know how much I turned the device), I get a reprducable error. It seems not to be influnced by calibration and looks like this: [image: grafik] https://user-images.githubusercontent.com/5088496/43047552-08047406-8dd9-11e8-8bb7-b176b876ed8f.png

Around 0° it is roughly 10%, but increases towards 180°.

Any suggestions? The result is the same when I use the MadgwickQuaternion calc. In the Mahony/Madgwick routines are some free parameters Ki and Kp. Should I change them? Or GyroMeasError/~Drift?

Thanks for your work! Joachim

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/kriswiner/MPU9250/issues/298, or mute the thread https://github.com/notifications/unsubscribe-auth/AGY1qtYZcnsk7oyeER73sAF8RvglJivsks5uJKMfgaJpZM4VaDL3 .

jhergeth commented 6 years ago

Thanks for the fast answer. I will try that.