if we integrate each axis gyro output independently at its sample time, how can we calculate maximum gyro total integration error for 1 hr from its data sheet parameters(ex:in-run bias stabilty,random walk etc), when it is kept at stationary.

We have purchased ADIS16475-2(gyro has in-run bias stability 2.5 deg/hr and random walk = 0.15 deg/sqrt(hr) ). After power on ,each gyro axis offset is calculated for 10 sec .When we integrated each axis gyro output(Offset compensated) independently at sample time of 10msec,the integrated theta is drifted as follows

X-axis integrated theta = -20 deg/hr

Y-axis integrated theta = 102 deg/hr

Z-axis integrated theta = 72 deg/hr

but in data sheet, bias stability is given as 2.5 deg/hr. But above test results(done at stationary condition) are too far from bias stability given data sheet.

Please explain me,what may be reason for too much difference between datasheet and Test results of above?

We have purchased ADIS16475-2(gyro has in-run bias stability 2.5 deg/hr and random walk = 0.15 deg/sqrt(hr) ). After power on ,each gyro axis offset is calculated for 10 sec .When we integrated each axis gyro output(Offset compensated) independently at sample time of 10msec,the integrated theta is drifted as follows

X-axis integrated theta = -20 deg/hr

Y-axis integrated theta = 102 deg/hr

Z-axis integrated theta = 72 deg/hr

but in data sheet, bias stability is given as 2.5 deg/hr. But above test results(done at stationary condition) are too far from bias stability given data sheet.

Please explain me,what may be reason for too much difference between datasheet and Test results of above?

The In-run bias stability (IRBS), is the lowest point of the Allan Deviation curve.

You may have to capture a longer set of data for you to acquire the IRBS. In this case, you would have to integrate around 100-second samples as opposed to 10msec.