Meaning of non-linearity in datasheet?

[Gideon]

Hello,

Could you explain the meaning of "non-linearity" in datasheet of BMI160?

Thanks,

[Vincent]

When we convert the register value to physical result,  we expect the output is linear.

Take example:  the corolation between physical acceleration and the MEMS output is treat as linear.    Then we apply 1 g on sensor,  sensor output is also 1g.   And this sensitivity efficient is identical in full range. 

This "non-linear" parameter indicate that the mapping of physical value as sensor output is not exactly fully linear.   so the sensitivity values might different in some point of accelration value within the range.   (just for example of explain:  0.5g physical accleration might output as 0.6g)

So when you estimate the sensor output error,  you need to consider this non linearity error inside. 

[Gideon]

Hi Vincent,

Thanks for your explanation.

but still, I don't understand How to calculate the non-linearity error limitation (+- 0.5%FS) in 2g range.

Could you let me know the non-linearity if I get the output as below?

For example,

reference value : 0.5g  / output : 0.52g

reference value : 1g  / output : 1.03g

reference value : 1.5g  / output : 1.43g

reference value : 1.8g  / output : 1.72g

[Vincent · Accepted solution]

Non linear calculation choose the biggest deviation in the range. 

Means in certain point the output has +-0.5% deviation. 

So when estimate the total measurement error in particular point, we can add this error.  

So the error can be 1g *(1+ / - 0.005) = 1.005 / 0.995 g output. 

[Gideon]

Hi Vincent,

Thanks for your explanation.

I have one more question but I am not sure you can share the information.

How many points(ex, 0.3g, 0.4g, 1g, 1.5g, 2g) you test to see non-linearity error(±0.5%FS)?