9150 Magnetometer readings not accurate or useable

Hello,

I have ported the Embedded Motion Driver 5.1 to our ARM based platform. When attempting to use the DMP and the Quaternion code provided to calculate the "fused" Euler angles I get completely useless values for the yaw.

When starting up the board, reagardless of physical orientation, the yaw value starts at around 0 and then counts up to ~-8. Every time... It makes no sense to me. When using YAW for heading, 0 Degrees should be magnetic north, correct?

I am attempting to use this as a heading but its not a heading right now. It appears that it is orienting to itself and its original position rather then the position of the earth's magnetic fields. Is this expected behavior?

Also,

I tried to use the raw magnetometer data to calculate the euler angles but it is so incredibly noisy that its sort of unusable.

Anyone have any insight to these problems?

Thanks!

Nick McCarty