Why the Z axis is noisier than the others axis?

[AJura.1]

Hi, I'm testing with the LIS3DHTR and I notice that the Z axis is noisier than the others axis. It doesn't matter the position of the accelerometer, always the Z axis is noisier. Why does happen?

[Eleon BORLINI]

Hi @AJura.1​ ,

the noise should be however always inside the datasheet specification for the LIS3DH (at p.11), i.e. 220μg/√Hz in the FS 2g case, isn't it?

You should also separate the offset DC contribution to the proper rms noise effect.

On the LIS3DHTR device family (which is not the last generation devices), the Z axis is by mechanical construction a little different from the x and y ones, since it is "out-of-plane".

You could try lowering the ODR to get a natural low pass filter on the integral (i.e. in the whole band) noise, since the acquisition band will reduce to ODR/2.

-Eleon

[AJura.1]

Hi @Eleon BORLINI​ ,

Yes, I know. I'm already separating the DC component from the offset but I still see a lot of noise. I share some small examples of the measurements that I have.

As you can see, the noise on the Z axis is very high. In the other axes the noise is also high, but acceptable. I am using a full scale of 2g, with an ODR of 100 Hz. I really don't know why only the Z axis is so noisy.

 

[Eleon BORLINI]

Hi @AJura.1​ ,

I see, thank you for sharing the data... you said that you are seeing the issue also when the device is in a vertical position, i.e. with the z axis not orthogonal to the ground, right?

Can you try lowering the ODR and checking the noise?

And did you have the possibility to test another LIS3DHTR?

By the way, to check if the device is good (maybe has been damaged in the past for some handling reason), you can run the self test procedure: it is described in the AN3308 application note p.57, and the C code for the application is on Github (lis3dh_self_test.c).

-Eleon

[AJura.1]

Hi @Eleon BORLINI​ ,

I have 3 LIS3DHTR more, I test all of them and I see the same results. In all of them the Z axis is noisier than the others. If I decrease the ODR I can see how the noise is lower but in the z axis is still very high. I use the 2G full scale and the High Resolution mode.

The results of the selftest are te same, a lower noise but still high. Could it be due to soldering process?

[Eleon BORLINI]

Hi @AJura.1​ ,

well, it could be...

Is your LIS3DHTR mounted on a custom board, right? I mean, you are not using an ST adapter such as the STEVAL-MKI105V, or other ST boards, to exclude handling or soldering process related issues.

Otherwise, is your soldering profile / process compliant with the JEDEC J-STD-020 specs (datasheet p.21)?

-Eleon

[AJura.1]

Hi @Eleon BORLINI​ ,

Yes, I use a custom board. Well, I read the JEDEC J-STD-020 specs and I think that could be the problem because we didn't bake for soldering during 24 hours. We only used a few minutes.

But it is very strange because if the soldering process is not exactly the correct the 3 axis would be noisy, but only the Z axis? I did the FFT of the signal and it seems to be white noise.

[Eleon BORLINI · Accepted solution]

Hi @AJura.1​ ,

the MEMS structure on the LIS3DHTR accelerometer is a little asymmetric with respect to the comb-finger structure along X and Y.

This may result in a stronger impact of bending / uniform soldering effects during the soldering process, this resulting in a noisier output along Z axis.

-Eleon

[AJura.1]

Hi @Eleon BORLINI​ ,

Ok, I understand. Is there a document where explains the internal structure of the accelerometer?

[Eleon BORLINI]

Hi @AJura.1​ ,

well, we cannot disclose the internals structure of LIS3DH MEMS accelerometer, but I can see online a reverse on this product, with some SEM picture of the MEMS chip, that may help you in better understanding the structure (link).

 

-Eleon