ADIS16497 gyro bias instability after power-on


We use ADIS16497-1 IMU in inertia navigation project.

But I can observe the following....

After power-on from a cold condition (internal temp about 22 C) all gyro outputs exponently changes from its initial values (about -0,13 deg/s for X, 0,18 deg/s for Y, -0,05 deg/s for Z) to target stable values (about 0,02 deg/s). This process take a time within a 15 mins. Internal temp of IMU changes for 4 degs. All accelerometers works perfectly.

After hardware reset of IMU all gyro outputs saves a stable values.

After power restart (10 sec pause) gyro outputs changes from a 0,01 to stable values.

After warm start (15 min pause in off condition) gyro outputs runs from a 0,1 - 0,15 deg/s to stable values. Internal temp changes from 28 to 33 deg C.

It's worth noting that axis Z has a better temp stability and has a minimal power-up changes.

Transient power voltage of IMU was within 20 mV. Temperature instabilities (for stable condition of gyro bias) was measured in temperature band from 15 to 65 C and not exceed values of 0,02 deg/s.

In datasheet rev.D for ADIS16497 Power-On Start-Up Time is a 265 ms and not include some aspects, which can affect overall accuracy.

Please say something more about start-up process in gyro. How long time needs for initial gyro stabilization process? What is the cause of the instability of the initial zero offset of gyro at power-up time?

Best regards,


  • 0
    •  Analog Employees 
    on Jun 13, 2021 2:40 AM

    Hello Andrew,

    I am sorry about our slow response to your request.  The short answer is that the gyroscopes do have a behavior, which causes drift at turn-on.  For this series of IMUs, the behavior typically follows a first-order, exponential decay formula: 

    Bias = Initial Bias Error x exp(-time/Tau), where the Initial Bias Error is typically 0.1-0.15dps and the time constant is in the region of 5-20 minutes. 

    I hope that this helps.  

    Best regards,

  • Hello Mark,

    thanks for your reply.

    You wrote earlier about exponential decay and we use to compensate this inttability an exponential curve.

    We continue to wait for a detailed description of the cause of instability what is essential for the
    correct choice of the compensation method.

    Best regards,


  • 0
    •  Analog Employees 
    on Jun 17, 2021 12:39 PM in reply to Andrew-777

    Thank you, Andrey!

    I am glad that you found useful information on this.  

    With respect to this question, we unable to discuss product design details, in public forums, but I can share that this behavior is inherent to the core sensor design. 

    • We continue to wait for a detailed description of the cause of instability what is essential for the
      correct choice of the compensation method.

    Generally speaking, I can offer the following, general observations: 

    • Initial error and time constant have some temperature dependence
      • At cold, the time constant is slower and the peak is lower
      • At hot, the time constant is faster and the peak is higher
    • Behavior appears to be repeatable
    • Behavior is lot dependent
      • In some lots, this behavior is non existen

    Are there any specific details that you need, with respect to the behaviors? 

  • Hello Mark,

    I agree, the discussion of the process is outside the public forum.

    We have now obtained the required measurement stability.

    Earlier, I sent a service request and a detailed description. I am sure the history of the correspondence has been preserved.

    I would be happy to continue discussing the methods of compensation outside the public forum.

    Best regards,


Reply Children
No Data