LSM6DSL Drift Specifications

[RBurg.2]

Hello, ST engineers. I am working with a general purpose IMU for a tilt sensor application. I am observing some drift on the sensor while the device is off. I have heard that all IMU's can drift a bit, but I was under the impression that it would happen over a longer period of time. Do you have any specifications or testing data regarding the drift of this IMU over time?

I am observing an offset drift of up to 3mg on some units just overnight. The readings are taken near 0g, but not exactly due to zero-g offset in device. The sensor is temperature calibrated, and the offset drift is measured near the same room temperature to minimize the effect of temperature related drift.

I would also appreciate anyone else who has input regarding if you think this magnitude of drift is reasonable and expected for this sensor?

Thank you,

Russell

[Federica Bossi]

Hi @RBurg.2​ ,

Can you give me some more information about your test? For example what settings do you have of Full scale and ODR? Are you in Low power mode or in High performance mode?

In addition, have you logged the temperature sensor output? Can you share it?

[RBurg.2]

Hi Federica, Thank you for the reply.

I am using the sensor in high performance mode, 2g scale, 104Hz ODR. I do read the temperature sensor output and use it for thermal offset compensation. I don't have a log of it right now, but I can tell you that it has about 0.5 degrees of noise on the reading so I filter it a bit. There are some nearby components that contribute to thermal noise, nothing extreme though, just a processor and a 100mW boost converter.

I am currently setting up a test to separate the IMU onto its own board. I can share results with you in a couple of days. And I will generate some logs of the temperature data before and after.

Best Regards

[RBurg.2]

After soldering the IMU on its own board, rigidly attaching it to the same enclosure as the main board. The temperature noise seems to have increased, and the overnight drift did not improve. Some axes had significant drift, some did not, similar results as other dev devices. I ran this test on 3 separate units.