H3LIS331HDL Z Axis reading high in static environment

BTatt.1 · ‎2021-12-22

Hello All,

We're having some trouble with our H3LIS331HDL and I'm hoping someone can point us in the right direction.

Below is a recording from a device with the high-G issue on the Z-axis. The device is in a static position on a desk and we're seeing the Z-axis with values of up to 4G. If we change the orientation of the device, the Z-axis still has this high G value.

This is what we're expecting to see - a noise range of less than 2G on all axis;

Now for the confusing part - this value shifts over time. We are seeing devices with no G offset issue that passed QA testing have this issue after 3 months of just sitting on a shelf. We're also seeing devices that failed QA at 3G noise now have 6G noise after 3 months.

The high G value isn't just when it is static, the unit when moved in a controlled manner shows the Z-axis reading higher than it should in comparison to a device without the high G issue.

QUESTIONS

What could cause a single axis to have a continuously high noise value?

Is there a way of resolving this through the firmware or is it a hardware issue?

We've changed manufacturers in case it was an assembly issue to no avail, moved the position of the accel on the board with no luck.

Any help anyone can offer would be truly appreciated as I'm stuck on this one.

Cheers,

Ben

Eleon BORLINI · ‎2021-12-23

Hi @BTatt.1 ,

does the issue appear after a mechanical shock (above the devices capabilities), harsh environmental conditions or just randomly?

Since the H3LIS331HDL doesn't unfortunately have the self test feature (that applies to MEMS stator and rotor a known electrostatic voltage), you could try to power OFF and power ON again the chip and check if the issue is still there.

If not, I suggest you to check if the device recovers after a mechanical shock or after a bake condition.

Finally, please note that in the datasheet the Typical zero-g level offset accuracy is defined as +-1g, without upper and lower limits.

-Eleon

BTatt.1 · ‎2021-12-23

Thanks for the quick response @Eleon BORLINI

This issue arises on brand new hardware straight off the assembly line. We do an end of line test and pick up the problem 95% of the time at this stage.

The other 5% are devices that pass initial testing and get assembled into a unit but get caught at other QA steps. Once assembled into devices we have no way of power cycling the accelerometer as it is connected to Vbatt, so once the initial boot occurs when the battery is connected it doesn't get powered down until the battery dies and the whole device is no longer of use. This could be 6-12 months.

The heat profile of the oven during assembly has been monitored tightly to make sure it's within guidelines.

the PCB is double sided, could something on the opposite side of the board be affected the z axis? We didn't think so due to some devices being fine and others not but worth asking the question.

does the H3LIS331HDL have any kind of internal calibration we could trigger when we know it's in a fixed position to clear this noise off?

Cheers

Ben