LSM6DS33 => problem with self test

[Maciek Zawistowski]

Posted on November 13, 2017 at 11:16

Hello,

I recently bought two Pololu IMU03a systems, and both of them gives me negative self test (both accel and gyro). To communication I use I2c protocol.  During the test the device is stationary

Calibration:

Rej. 0x10h => Value: 0x30h;

Rej. 0x11h => Value: 0x5Ch;

Rej. 0x12h => Value: 0x44h;

Rej. 0x13h => Value: 0x00h;

Rej. 0x14h => Value: 0x00h;

Rej. 0x15h => Value: 0x00h;

Rej. 0x16h => Value: 0x00h;

Rej. 0x17h => Value: 0x00h;

Rej. 0x18h => Value: 0x38h;

Rej. 0x19h => Value: 0x38h;

After calibration i wait 200ms for stable output, then i erase first five values (by reading registers from 0x22h to 0x2Dh).

Then I read these five registers, I combine them (0x22h with 0x23h ...) and average them in each axis (OUT_NOST).

Then I write 0x01h into registry 0x14h to enable self-test, wait for about 200ms, and discarded next five values (from registers from 0x22h to 0x2Dh).

Then I read five times values for each axis (OUT_ST), calculate average those values, and find min and max value for each axis.

For each axis I make a compare as below:

abs(min_ST_x)<=abs(OUTX_ST - OUTX_NOSTD)<=abs(max_ST_x)

Values I recive:

XGyro without self test [8, 14, 6, 9, 7], with self test [ 2994, 4491, 4499, 4492, 4494]

YGyro without self test [-66, -74, -66, -62, -73], with self test [ 2649, 4020, 4020, 4014, 4

015

]

ZGyro without self test [-1, -1

, 

-1

, -1

, -1

], with self test [ 3207, 4848, 4843, 4845, 4845]

XAkcel without self test [575, 505, 455, 502, 564], with self test [ 8114, 7962, 8036, 8056, 8098]

YAkcel without self test [434, 445, 529, 480, 449], with self test [ 9674, 9734, 9681, 9732, 9668]

ZAkcel without self test [

-16131

,

-16276

,

-16273

,

-16260

,

-16154

]

, with self test [

 -10374

,

-10543

,

-10510

,

-10446

,

-10412

]

Am I doing something wrong, or should I return those IMU's ??

Thanks for any answer or clue

[Petr S]

Posted on November 13, 2017 at 17:50

Hello Maciek,

I suggest trying to split the accelero and gyro self-tests instead of measuring them at the same time. Also I'd suggest to make the register settings exactly as defined in

http://www.st.com/content/ccc/resource/technical/document/application_note/9c/d9/07/d0/d4/a9/45/00/DM00175930.pdf/files/DM00175930.pdf/jcr:content/translations/en.DM00175930.pdf

, that's mandatory for correct self-test. See if it helps.

You can also download the 

http://www.st.com/content/st_com/en/products/embedded-software/mcus-embedded-software/stm32-embedded-software/stm32cube-embedded-software-expansion/x-cube-mems-xt1.html#getsoftware-scroll

 SW package where you can find similar SelfTest example. This package supports a few Nucleo and MEMS Shield boards but if you don't have them you can still at least inspire yourself by the open source code.

Best regards,

Petr

[Maciek Zawistowski]

Posted on November 14, 2017 at 07:39

Hi,

thank you for your response, I tried your approach as in

/external-link.jspa?url=http%3A%2F%2Fwww.st.com%2Fcontent%2Fccc%2Fresource%2Ftechnical%2Fdocument%2Fapplication_note%2F9c%2Fd9%2F07%2Fd0%2Fd4%2Fa9%2F45%2F00%2FDM001759pdf%2Ffiles%2FDM001759pdf%2Fjcr%3Acontent%2Ftranslations%2Fen.DM001759pdf

p.77, Values I get are approximately the same, and both akcel and gyro self test fail. I think I need to send back those devices. Why does this issue happened, the device is brand new (yesterday I've unpack it). Is there a possibility that device get broken during transport? Is those devices so fragile?

[Petr S]

Posted on November 14, 2017 at 10:33

Hi Maciek,

I've made a quick calculation for X axis of acc and gyro and the self-tests passed.

For linear acceleration (example of X axis):

Linear acceleration sensitivity for FS = ±2 is 0.061 mg/LSBAverage of 575, 505, 455, 502, 564 is approx. 520 LSBAverage of 8114, 7962, 8036, 8056, 8098 is approx. 8053 LSB8053 - 520 = 7533 LSB * 0.061 mg/LSB = ~460 mgUse these self-test limits:90 <= 460 <= 1700 mg => passed

For angular rate (example of X axis):

Angular rate sensitivity for FS = ±2000 is 70 mdps/LSBAverage of 8, 14, 6, 9, 7 is approx. 9 LSBAverage of 2994, 4491, 4499, 4492, 4494 is approx. 4194 LSB4194 - 9 = 4185 LSB * 70 mdps/LSB = ~293 dps

Use these

self-test limits

:

150 <= 293 <= 700 dps => passed

You can calculate remaining axes but I believe it will pass too so I wouldn't return the devices.

Best regards,

Petr

[Maciek Zawistowski]

Posted on November 14, 2017 at 12:11

Hi,

thank you very much for your response. In AN4682 there is for XAxis:

| Min (ST_X) | <= | OUTX_ST-OUTX_NOST | <= | Max (ST_X) |.Could you tell me how do you have calculate this line:90 <= 460 <= 1700 mg => passedI've trouble, I do not know how 90 value is calculated.

In my opinion it should be 7962 LSB * 0.061 mg / LSB ~ = 486mg (min value during test condition)

[Petr S]

Posted on November 14, 2017 at 12:31

Now I finally understood what was the issue here. To explain it, the Min and Max values in

| Min (ST_X) | <= | OUTX_ST-OUTX_NOST | <= | Max (ST_X) | formula 

are actually the self-test limits given by ST. So you can rewrite your formulas as follows:

For linear acceleration:

90 mg <= | OUTX_ST [mg] - OUTX_NOST [mg] | <= 1700 mg

90 mg <= | OUTY_ST [mg] - OUTY_NOST [mg] | <= 1700 mg

90 mg <= | OUTZ_ST [mg] - OUTZ_NOST [mg] | <= 1700 mg

For angular rate:

150 dps 

<= | OUTX_ST [dps] - OUTX_NOST

[dps] | <=

 700 dps

150

dps 

<= | OUTY_ST [dps] - OUTY_NOST

[dps] | <=

 700 dps

150

dps 

<= | OUTZ_ST [dps] - OUTZ_NOST

[dps] | <=

 700 dps

If your calculated values fit into the interval, the device is ok.

Best regards,

Petr

[Maciek Zawistowski]

Posted on November 14, 2017 at 12:53

And one last question where can I find those limits, are they depend for scale which I use, or are they constant??

[Petr S]

Posted on November 14, 2017 at 14:18

At this moment you cannot find it anywhere, just use what I gave you, please. It's FS dependent so you have to strictly follow the procedure from application note.