roll affecting yaw?

[hatt0002]

Hi friends,

I have managed to read from two mpu6050 on DMP. Both sensors are fitted on a piece of board and are perfectly levelled. 
My problem is that when I tilt the sensors such that I increase/decrease Roll, the Yaw is changed, even when the sensors are not rotated on the z axis. Is it normal that rolling affects yawing? Furthermore, the change in Yaw for both sensors is not the same, and there will be a difference in yaw of 8 deg between the two sensors for a 45deg rolling.

Thanks in advance,
Hatt

[Arcfour]

Same strange problem here.

 

I had also another issue with interrupts and removed

while (!mpuInterrupt && fifoCount < packetSize) {}

in the sample file to get it running, because even after minutes no interrupt appeared. But afaik this have nothing to do with the roll/yaw problem?! Im using a Teensy 3.1 board.

[radleyp]

Hi Hatt,

 

I saw your post back in 2014 about integrating two MPU6050. I am doing similar project right now to track the arm movement. I also encounter FIFO overflow problem. Would you mind sharing the code to fix the issue ? Thank you.

[zenman]

I am experiencing the same thing.

I don’t have a simple solution, but I have basically tracled it down to the yaw/pitch/roll algorithm.

What I have found is that if you stick purely with quaternions and do not switch to yaw pitch roll, the flipping problem goes away. At least it does in the processing “teapot” simulator.

My current issue is that I just don’t understand how to work with quaternions very well. Is anyone here well versed in quaternions?