robjames66

The mag on the 9150 is fine as long as you're happy filtering it. My HMC5883L gives be a heading +/- half a degree and with the 9150 I get +/- 2 degrees. Proper hard/soft calibration helps a lot and a hefty low pass filter.

It sucks doesn't it! Much worse than a HMC5883 A simple low pass filter helps. If x,y hold the raw values and xf,yf are global floats: xf = xf - (xf / 24) + (float)x/180.0; yf = yf - (yf / 24) + (float)y/180.0; // not tilt compensated... heading = atan2(xf / 24, yf / 24);

I used the Motion Driver 5.1.2 core libs and wrote a test sketch using the the motion driver test example code.

I've ported by application from the I2CDevLib MPU6050 libraries to the official Invensense libs and I get very poor DMP behaviour. The reported orientation does not always return to the original value when there us yaw rotation one way and back. This is not drift - it's a sudden step change in the yaw and it sucks. The I2CDevLibs behave much better for me. Invensense forums are a wast of time (full of spam) and very little response to queries. Rob J.

Hi , Does temp effect drift in DMP output ?

There are other causes of drift but this is one way to address low levels of background yaw drift without the need for magnetometers. Works very well for me. http://reprapdad.wordpress.com/2014/04/16/mpu-6050-yaw-drift-and-how-to-combat/ Cheers, Rob J.

Hi Folks, I've been using the I2C MPU-6050 libs as a basis for an arduino based orientation tracker. All working well but I'm now delving deeping into the Invensense docs and examples and they differ in how calibration is handled. Do I have the following right ? The gyro/accel offset registers used by the I2Cdevlibs (setXGyroOffset etc) appear to be undocumented by invensense and used for factory calibration. I can see these are useful if re-calibration is needed as they're non-volatile and automatically used by DMP. The invensense motion driver code/examples recalibrate by pushing bias values calulated by the host (from raw gyro/accel data) down to the DMP using dmp_set_gyro_bias, etc. This is volatile so needs to be stored by the host somewhere such as additional EEPROM. The DMP can also be configured to perform 'self calibration' of gyro data after 8 seconds of no motion. Using the I2Cdevlib calibration I get good results. To combat yaw drift I add a drift compensation to the values which works very well as yaw drift seems to be consistent and stable even with lots of motions. Cheers, Rob