About accelerometer LIS3DH

[OHide.1]

I am considering an accelerometer.

The accelerometer is experimenting with LIS3DH.

I want to mount an acceleration sensor on a surgical light and 

detect the movement of the lighting part of the surgical light.

I want to issue an INT interrupt when the light section is moved.

The light section hangs from the arm as shown in the photo linked below.

The movement is relatively slow and smooth.

https://www.skylux.co.jp/product/ixm_cj/

Currently, I am using Raspberry Pi 4 and conducting a motion detection 

experiment with the LIS3DH evaluation program.

I'm in trouble because I can't detect slow movements.

When the X, Y, Z values ​​were monitored, 

the slow horizontal movement (about 7.5 cm / sec) was as shown in the attachment, 

the value is buried in noise and the change cannot be detected.

Is it difficult with LIS3DH to detect relatively slow movements?

In addition, I would like to consider IIS2ICLX, IIS3DHHC, and IIS2DLPC, 

which have relatively low noise levels.

Is it difficult to move slowly?

I would appreciate it if you could give me some advice.

Thank you.

Best Regards

[Eleon BORLINI]

Hi @�?野 (Ono) 英尚 (Hidenao)​ ,

let me first notice that this is an interesting application.

So your focus is to get high accuracy for slow (quasi static) movements? Which is -indicatively- the maximum frequency you want to detect? I believe less than 25Hz, but it depends exactly on the application, and on the signal you want to monitor (manual movement of the surgical light?)

Suppose to use the LIS3DH that you already have, my suggestions is to configure a low ODR (e.g. 10Hz) and low FS (+-2g), or to use a mid range ODR (50Hz /100Hz) but with the HP filter enabled. Try also to disable the BDU, if you enabled it.

If you can change the device, I agree with you that the more recent IIS2ICLX, IIS3DHHC and IIS2DLPC devices are more indicated for a small signal detection, having a lower noise and a 16-bit dynamic: in particular, with the tri-axial IIS3DHHC (iis3dhhc_read_data_polling.c for the drivers) you can set lower FSs and/or a Low pass filter, you have ultra-low noise performance (15 µg/√Hz). You could also use a LIS2DW12 accelerometer for a good compromise, which has appreciable performances (90 µg/√Hz noise density and 16 bit resolution).

On top of the an important aspect to evaluate is to check the compatibility of the accelerometers with the a medical application such as the surgical light control...

-Eleon