Gyro data(Angular rate) Integration

The IMU ADIS16475-2 is kept at stationary (@ Constant temperature) data logged at rate of 100Hz into CSV file. If we integrate gyro X,Y and Z independently at rate of 0.01sec.the gyroscope integrated values after 1 hr as follows. The integration was done after 10 sec from power ON by compensating each gyro axis OFFSET. This OFFSET is calculated for 10sec from POWER ON.

X --> -125.5 deg

Y--> 44 deg

Z-->90 deg.

If we integrate gyro X,Y and Z independently at rate of 100sec i.e integration time dt is taken as 100 and integrated for every 100sec gyro sample. The gyroscope integrated values after 1 hr as follows. The integration was done after 100 sec from power ON by compensating each gyro axis OFFSET. This OFFSET is calculated for 100sec from POWER ON.

X --> -112 deg

Y--> 52deg

Z-->89deg.

In ADIS16475-2 datasheet Bias instability is given as 2.5 deg/hr but in practical it is drifted too much.

Please suggest , is the above results are valid or not?

  • 0
    •  Analog Employees 
    on Mar 23, 2021 7:42 PM


    Thank you, for your post.  These devices have a drift behavior, in their bias, which we refer to as "turn-on drift."  Since we have released this product, we have developed stronger understanding for this behavior.  In general, we can model this behavior, using a first-order, exponential decay, where the initial bias error is typically within  +/-0.75dps and the time constant is in the region of 5-20 minutes.  Here is an example of the model, when the initial bias error is 0.6dps and the time constant is 10 minutes. 

    Bias Error = 0.6 x exp (-time/10minutes)

    I hope that this helps! 

  • Thank you for answer

    1.You said that the bias is exponential decay, but Gyro-Z in ADIS 16475-2 found to be exponential rise, i.e at turn ON its value is 0.06dps after 1hr it's value reached to 0.09dps.Why?

    2.How to calculate the maximum drift angle/hour which is calculated by integrating gyro output(dps) at sample rate of 0.01sec in stationary conditions from ADIS IMU data sheet?

    3.How much deviation can  we expect in maximum drift angle/hour from stationary to dynamic conditions (Linear and Angular accelerations)?

    please respond to above queries. 

  • 0
    •  Analog Employees 
    on Apr 19, 2021 2:28 PM in reply to Deviprasad

    I am sorry for the lack of response on this. We lost a key staff member and are working to re-organize our team to provide quicker responses, in the future.  We appreciate your persistence. Please see my preliminary responses, below.  

    1.You said that the bias is exponential decay, but Gyro-Z in ADIS 16475-2 found to be exponential rise, i.e at turn ON its value is 0.06dps after 1hr it's value reached to 0.09dps.Why?

    NevadaMark-> In this case, it sounds like the model might be following this formula:

    0.06dps + 0.3dps x (1 - exp(-time/TC)), where TC is in the region of 15-25minutes.  Does this match the plot that you are observing?  If not, can you share the plot?   

    2.How to calculate the maximum drift angle/hour which is calculated by integrating gyro output(dps) at sample rate of 0.01sec in stationary conditions from ADIS IMU data sheet?

    NevadaMark -> The datasheet does not provide maximum values, nor can we offer them in this forum.  However, we can offer that these typical values are conservative, with intent of covering >70% of the parts being as good as the Typical or better. 

    3.How much deviation can  we expect in maximum drift angle/hour from stationary to dynamic conditions (Linear and Angular accelerations)?

    NevadaMark -> Same comment on "maximum" specs. Generally speaking, the metrics are setup to support a second order model estimation: kg x a + VRE x a^2.  What type of acceleration are you trying to quantify the impact for?