In my project I use the sensor  ADIS16480 to determine the pitch and roll during movement of a car. I need to know a value of an angle accurate to 1 degree. At the moment I use Euler Angles. The settings of the sensor   are used by default . While I move the sensor on a plane surface I can see, that the  value of the angle is 5 degrees, and when I move the sensor circle-wise during 3-4 seconds, the  value can be  changed to 15 degrees. I think, it`s an error of the sensor, because the value of an angle must not exceed 0.3 degree.  An application of AVERAGING/DECIMATION FILTER was useless ( the magnitudes DEC_RATE = 245 and 2047 were used)

What must be done to obtain an essential value of pitch and roll on conditions that a car is moving and  the frequency  of an inquiry of the sensor  is 10 hz.

Thank you.

  • 0
    •  Analog Employees 
    on Apr 22, 2015 4:29 PM

    Thank you for posting your question. First, the 0.3 deg has specific conditions tied to and it is a typical number, not a maximum.  Second, what has been done to validate your expectation of pitch/roll angle during these maneuvers?  Are you assuming that the angle stays at 5 degrees during the turn?  Or, are you measuring it with a higher-precision IMU for reference?

    Can you post data for all 9 sensors and your Euler angles during a single turn?  Perhaps there are clues that might help us understand what could be causing this.  If you are collecting data at 10Hz, we would suggest using 32-bit data and the DEC_RATE settings that produce the 10Hz update rate in the output data registers.

    We look forward to your response.

  • Thank you for you answer I would like to know under what conditions you obtain the precision 0,3 deg or 1 deg of pitch/roll ( the second is also satisfactory). Unfortunately, now I cannot give you data , which were taken from all sensors during the turn, because I don’t use your evaluation board. I used DEC_RATE at the frequency of 10Hz, but it did not help and I noticed an error, which can amount to 20 deg during long influence. 

    As for 5 deg, this value is observed single, not for a long time.

    Thank you.

  • Have you calibrated the accelerometers and magnetometers?  The data sheet says the specified accuracy for attitude assumes the sensors have been calibrated (see note 1 in Table 1).

    Also, in your bench top test, are there other electronics that could be causing magnetic field disturbances?  If you have the device sitting next to a computer, and you are moving the device in circles, the computer could be messing up the magnetometer readings. 

  • The calibrations where not  carried out, because I thought that the defaults will be enough. And influence of electromagnetic noise was minimized, so I don’t think it’s the  reason.

    Could you tell me,please, what calibrations are necessary and in what condition of the sensor?

    Thank you

  • 0
    •  Analog Employees 
    on May 2, 2015 11:54 PM

    I am sorry that  I have not been more available to help on this quickly.  The most common cause of this type of behavior has been through unexpected magnetic field interference. Have you done an XY plot to see what the hard and soft iron errors are?