• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 2 answers
  • Subscribers 46 subscribers
  • Views 4264 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA240: Nonlinearity error definition consult

EMMA ZHANG
Prodigy 50 points
Part Number: INA240


Hi team, 

May I know how is the spec item 'Nonlinearity error' defined and measured?

1) In my understanding, the nonlinearity error is calculated as follows:

    Firstly, we plot output versus input and do the linear regression to get a best-fit straight line.

    Secondly, we calculate the deviation of output from this best-fit line and this deviation is divided by best-fit line to transfer into percentage value

    Lastly, the maximum percentage value is the nonlinearity error.

    Is the nonlinearity error defined in this way?

2) The nonlinearity error got by the above steps is refer to the output. 

     Has the nonlinearity error in the spec item been referred to the input? Is the value divided by the actual Gain already?

3) How is this nonlinearity error tested? what is the sweep steps?

    And how can we guarantee that the maximum deviation is covered by the step?

Thanks for your answer in advance.

  • 0
    TI__Mastermind 35290 points

    Hi Emma

    1)     Yes, it is defined this way.

    2)     The data sheet typical spec is referred to input, in other words the gain is already taken into account.

    3)     This parameter is only characterized with a typical spec, and not 100% tested for each and evey device. That’s why there is not a max limit in the datasheet.

    Regards, Guang

  • 0 in reply to Guang Zhou
    Prodigy 50 points

    Hi Guang, 

    Thank you very much for your fast reply.

    I still have a question on the definition of nonlinearity error. 

    In you reply, you confirmed that the error is the max deviation from the best-fit line which is got by linear regression. But I find a TI training material on CSA in which the error is defined as the max deviation from a calibrated gain curve. The gain curve is calculated by linear approximation using two data points at 10% and 90% of the full-scale output range.

    Therefore I am confused. Which is the correct definition? How to get the 'straight line', by linear regression or by 2-data-points?

    The training material is as follows:

     

  • 0 in reply to EMMA ZHANG
    TI__Mastermind 35290 points

    Hi Emma,

    Here is my understanding regarding the inconsistencies:

    The best-fit line method is a more strict definition of nonlinearity. There are several simplified versions in use in order to reduce the amount of computation, especially during final test if such test is included.

    For example the simplest of these is a 3-point test. Based on observation/characterization, a third point with the most deviation is chosen and nonlinearity is calculated simply based on these three data points.

    It is a result of compromise between scientific rigor and engineering practicality.

    Regards, Guang

  • 0 in reply to Guang Zhou
    Prodigy 50 points

    Hi Guang,

    Thank you for your reply.