LIS3DH Z output noisy

[Anusuya Nallathambi]

Posted on June 05, 2017 at 11:40

I am trying to measure a machine's vibration using LIS3DH sensor. 

I set the LIS3DH in low power mode and 5.376 kHz sampling frequency which outputs 8 bit output. I use spi to get the data from the sensor. I changed this raw output into �g' using the sensitivity value given in the datasheet.

My problem is: The sensor shows a fluctuation in its output of Z axis even when there is no vibration applied and the sensor is fitted on a horizontal surface. The range of this fluctuation is approximately from 0.6g to 1.2g. I have attached a image that shows such output. (X axis is time and Y axis is vibration in terms of g)

Can someone please help me to interpret why this is happening ? Is there a method to somehow remove this noise from the output ?

Please help.

#lis3dh-spi #lis3dh #mems-accelerometer

[Miroslav BATEK]

Posted on June 05, 2017 at 15:42

The fluctuation is really big, it is not correct.

Can you please share your sensor configuration and the code how do you convert the raw data into acceleration in g?

[Anusuya Nallathambi]

Posted on June 06, 2017 at 02:27

Thank you for your reply. I do not understand what you mean by sensor configuration. I am using LIS3DH sensor in low power mode. The sampling frequency is 5.376 kHz and the output is 8 bit long.The range of output is +/- 4g. I use FIFO stream mode and watermark interrupt to accumulate the data. After reading the data from the FIFO I multiply it by 0.032 to convert it into g. I use Raspberry PI to process the data via SPI communication. I am unable to show you the code right now due to some technical difficulties.

[Alex Spurling]

Posted on July 03, 2017 at 10:33

Please try the following and see if you still get the same noisy output:

Verify that the x and the y axes do not show the same noisy output.

Set the sensitivity to 2g.

Set high precision mode on.

Set the sample rate to something like 10/50/100hz (I've tried these rates and they work well for me)

Turn off FIFO mode and try to sample the acceleration values directly from the 28-2D registers.

Try reading both the low and the high bits as a 16 bit value. You then need to change the scaling factors to get the correct g value.

I found this thread because I was seeing more noise in the z axis on my sensor than on the x or y. However, the maximum deviation from the mean is only about 0.05g (x and y tend to stay within about 0.025g).