# Looking for gyroscope drift spec in datasheet

The product page of the BMI088 IMU makes the very impressive claim

The automotive-proven gyroscope of the BMI088 has an unmatched bias stability of less than 2°/h

However, I can't find this claim anywhere in the datasheet. The datasheet does contain Zero-rate Offset, Zero-rate Offset Change over Temperature, and Zero-rate Offset Supply Voltage Drift, but not Zero-rate offset drift over time.

What would zero-rate offset drift over time be called in the datasheet? If it isn't included, can it be derrived from other specifications in the datasheet?

• Does "bias stability" equal drift in all relevant contexts? – Russell McMahon Sep 17 '19 at 9:41

BMI088 is offering a wide acceleration measurement range (up to 24 g), high vibration suppression ratio and vibration robustness, as well as high bias and temperature stability. The automotive-proven gyroscope of the BMI088 has an unmatched bias stability of less than 2°/h and a low temperature coefficient of offset (TCO) below 15 mdps/K.

I don't think I would believe that either unless listed in the datasheet. Only the temperature drift is listed in the datasheet. You would have to know the internal workings of their Kalman filter and how the errors of the gyroscope are being integrated to be able to calculate this yourself. The best course of action would be to contact Bosch and ask them about the drift rate and to put that figure in the datasheet.

I think you're being snowed.

The zero-bias offset temperature sensitivity is $$\0.015^\circ/\mathrm{s/K}\$$, and the supply voltage sensitivity is $$\0.1^\circ\mathrm{/s/V}\$$.

Simple math tells you that's $$\54^\circ/\mathrm{hour/K}\$$, and $$\360^\circ\mathrm{/hour/V}\$$.

So the only way to realize an "unprecidented" $$\1^\circ/\mathrm{hour}\$$ bias drift would be to ovenize the thing and firmly regulate its temperature (possibly by including the regulator in the oven).

• and that assumed ONLY the temperature and VDD changes ----- caused drift. – analogsystemsrf Sep 17 '19 at 1:50
• My usual limit is one absurdly tight specification -- I'm definitely going to stop at two. – TimWescott Sep 17 '19 at 15:10