Floating point numbers are a representation of the numbers that permits the under stood positon of the decimal point to move or "float" around as necessary so that the actual full overall magnitude of the number is realized whilest keeping track of the number of with a set amount of precision. Now I say "precision" somewhat loosely here because in actuality we are dealing with keeping track of set amount of significance of the number. You can see that the equivalent significant number of digits out of an example mantissa of 24 bits implies quite a different absolute precision when carrying around an exponent of 10^^12 as compared to having an exponent of 10^^-12.

Floating point numbers make it easy to take something simple like converting millimeters to inches,,,,i.e. you just simply divide by 25.4 and the result is right there for you in the floating point result out to more decimal places than you probably even care about.. Whereas in the native integer format of the microcontroller it is necessary to "manage" or plan where the decimal point is at all times. One could say as such that the decimal point is "fixed" at a particular spot.

Mike Karas