I'm using an LSM9DS0 IMU and looking at the data sheet here: www.st.com/resource/en/datasheet/lsm9ds0.pdf.

In table 3 p13, it lists the sensitivities of the 3 main measurements in:

  • mg/LSB "milli gees per least significant bit"
  • mgauss/LSB "milli gauss per least significant bit"
  • mdps/digit "milli degrees per second per digit"

What does "per digit" mean?

There are good answers already about the meaning of "per least significant bit."

Looking at other places where this component is used is seems to mean nothing different from "per LSB". I.e., "per count of the encoder" But why is it spelled differently?

  • \$\begingroup\$ Well, bit is a kind of digit (more specifically, a binary digit), and a single digit has the same value regardless of the base you use, so it should be the same. \$\endgroup\$ Commented Mar 16, 2017 at 14:25

2 Answers 2


The g means the acceleration of gravity, 9.8 m/s^2. You can set 5 ranges. The highest range is the least sensitive at 0.732*9.8 m/s^2 for each LSB. There are 4 ranges of magnetic measurement. You pick the one you want, plus or minus 2,4,8, or 12 Gauss. Once you pick the range, the sensitivity tells you how magnetic flux density it take to increase the count by one. For the 12 Gauss range it takes .48mGauss to change the output by one. There are three ranges of angular measurement. The highest is 2000 degrees/second. When set to this range it takes 0.070 degrees per second to change the output by 1 count.

  • \$\begingroup\$ So, to answer the question specifically you are saying there is no difference between per LSB and per digit? \$\endgroup\$
    – tahsmith
    Commented Nov 4, 2016 at 23:46
  • \$\begingroup\$ I think so. Here is how I read this. Sensitivity is the range divided by 32768. If the 2000 degree/second range is selected it should take 61 mdps to change 1 bit. But the specification is 70 mdps. It seems they have combined the errors into the specification by saying a 70 mdps change is guaranteed to change the output 1 bit. It is interesting that except for the highest acceleration range the sensitivity is exactly the range divided by 32768 to three digits. That is impressive. So yes, I believe all the sensitivity numbers are the change that will guarantee a 1 bit change. \$\endgroup\$
    – owg60
    Commented Nov 5, 2016 at 1:00

A numerical digit may be a binary, octal, decimal or hexadecimal digit, see here https://en.wikipedia.org/wiki/Numerical_digit. There is no definition in the datasheet what kind of digit is used here.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.