LIS2DE12 self test

[RKeit.1]

I am trying to set up a self test process for an assembly including an LIS2DE12. The device works normally when I tilt and invert it. When I try to put it in to self-test mode by writing either 02h or 04h to CTRL_REG4 the readings are the same as with 00h written to CTRL_REG4.

I initially tried using the process for the LIS3DH on Page 57 of AN3308 but that says to write 80h to CTRL_REG4, which does not seem to be valid because that would set the BDU bit to 1, not valid for an 8 bit part, and not set either of the self test bits.

Can someone please shed some light on whether the self-test function is available with the LIS2DE12 and what I can expect from it.

Thanks in advance,

Rob

[Eleon BORLINI]

Hi Rob @RKeit.1​ ,

I suggest you to check also the C code implementing the self-test procedure for the LIS2DE12 that you can find on Github at this link --> lis2de12_self_test.c

This procedure is a little long, so it might be useful to have a double check on that code, besides the procedure described in the AN3308 p.57 (which is for the LIS3DH )

Please let me know if the code is useful for your purposes.

-Eleon

[RKeit.1]

Hello Eleon,

Thank you for your answer. That gives me some clues. One thing I am still unsure of is the magnitude of the variations in readings during self test. Is it like a few LSBs or bigger? Does it occur on all axes? I know the answers are probably in the documentation but I have winter brain fog.

I see the term mg used. Is that milli-gravity? If so then it could be too small for the readings to be noticeable with only 8 bits or buried in the noise from my wobbly desk. I am working with an 8 bit MCU to develop the test process so I am just looking at the raw X, Y and Z bytes.

Also, does the device stay in self-test mode indefinitely, until CTRL_REG4 is re-written?

Rob

[RKeit.1 · Accepted solution]

Hello again Eleon,

I have found the problem and now getting good consistent results in self test mode. I had forgotten to set the write bit used by the state engine that controls the SMBUS in the MCU I am using. I still can't get it to work with the Arduino but that doesn't matter.

Results I am getting, with the device sitting at an angle:

$ normal mode CTRL_REG4 = 00h

FD 19 3B                                    

FE 19 3A                                    

FF 19 3A                                    

FE 18 3A                                    

FE 18 3B                                    

!                     self test mode  CTRL_REG4 = 04h                

E1 FD 51                                    

E1 FB 50                                    

E2 FD 4F                                    

E2 FC 51                                    

E2 FC 50                                    

E3 FD 50                                    

$ normal mode CTRL_REG4 = 00h                                       

FE 19 39                                    

FE 18 3A                                    

FE 19 3A

Thanks again,

Rob

[Eleon BORLINI]

Very good @RKeit.1​ !

Remember also to use the LSB-to-milli "g" (meaning 1 gravity acceleration 9.81m/s^2 divided by 1000) conversion formulas. This is the standard magnitude for computing the acceleration with an accelerometer. The formula in the FS = 2g case is (source: lis2de12_reg.c)

float_t lis2de12_from_fs2_to_mg(int16_t lsb)
{
  return ( (float_t)lsb / 256.0f ) * 15.6f;
}

-Eleon

[RKeit.1]

I do all that on the cloud. I am working with a very basic MCU and do everything in raw bytes.

Also. Not in the documentation: There are two test modes. Mode 1, CTRL_REG4 = 04, moves the readings one way and Mode 2, CTRL_REG4 - 02, moves them the same amount in the opposite direction. Average the readings from both modes. If the answer is the regular reading (within two LSBs) then all is OK.

Now I am having fun (not) trying to reliably generate interrupts at 10 Hz ODR.

Rob

[Eleon BORLINI]

Hi Rob @RKeit.1​ ,

I'll go on your other post for this topic.

-Eleon