LSM6DSRX - Repeatable noise on X axis

[Candrian]

Hello,

I'm working on a project where I use the LSM6DSRX. I have noticed that the Gyro has a strange behavior, reporting peaks on the X axis while the device is solid static. The peaks may be correlated to the temperature change but it's not clear yet. I did run a test, on ambient temperature range +5C to +30 with two end-devices (DUT1 & DUT2) and one of the two reported the peaks on X axis at around 33C (internal temperature) while the other not. Is this something you have ever experienced or reported by any user? Since now all of my experiments shows that it is device related and not system related. I have replaced the LSM on some of the faulty and the issue seams to have been resolved. I have taken X-ray scans to verify that the soldering is acceptable and it is.

Following raw data as read from the sensors Gyro from 3 different devices. You may see the noise coming after a few minutes with a duration of around 50-60 minutes The samples are taken every second. The noise is always on X axis. 

A longer duration test around 5 Hours. 

Is this something any one has experienced? 

[Federica Bossi]

Hi @Candrian ,

No, it is not expected. Can you share your settings?

[Candrian]

Hi, @Federica Bossi  

Thank you for your prompt reply.

The settings are:

CTRL1_XL = 0x50
CTRL2_G = 0x5C
CTRL3_C = 0x44
CTRL10_C = 0x20  

The rest are defaults. 

[Candrian]

Hi @Federica Bossi ,

I noticed that SDx and SCx in our design are floating and we are operating the LSM in mode 1.

I patched one system, hard connecting SDx and SCx to GND using patch wire and it looks like the noise has gone. I can't confirm that this is absolutely related to this wire patch yet. Could this be the cause? 

Does SDx and SCx need a hard connection to GND or Vdd_IO or they can also be pull up/down with resistors? 

If this is the cause, is there any configuration we can set to the LSM to in order to prevent the patch wires (hard connection of SDx and SCx to GND or Vdd_IO)? This will require us to open all the devices using the LSM and perform the patch which is something we would like to avoid at this time. 

[Candrian]

@Federica Bossi,

My tests show that the SDx and SCx hard connection to GND does not fix the noise issue. 

I also tried with no positive results:

• Re initialization of the sensor while the noise presents. By initialization I mean re-setting the registers 

CTRL1_XL = 0x50
CTRL2_G = 0x5C
CTRL3_C = 0x44
CTRL10_C = 0x20  

• Move/Shake the device while the noise presents and leave it to rest on the desk. 
• Measured the ripple voltage on VDD and VDD_IO less than 1.5mV AC RMS
• Increasing the ODR from 208Hz to 6.66Khz results to high noise on all 3 axis! (Higher than the mentioned noise on X) 

If understand correctly from the datasheet Gyroscope LPF2 is always on? 

[Candrian]

@Federica Bossi,

If I read the sensor every 1mS via I2C, I can determine a noise duration of 4 to 8mS which repeats with a random pattern. It mostly ranges to negative values.