I'm sorry to beat this topic to death, but I still don't understand how
the integer-valued samples are scaled to floating point numbers.
If the scaling factor is 20/2^31, then I should be able to acquire SGL
values, multiply them by 2^31/20 (using double-precision floats to do
the calculation), and get integers.
When I do that on a bunch of datasets I've collected, I get fractional parts
that are multiples of 1/5, ie 0.2, 0.4, 0.6, 0.8 and 0.0
So the scaling factor cannot be 20/2^31.
By the same reasoning, it cannot be 20/(2^31-2^8) either, as the
*corrected* NI documentation states
here.
Sorry to be so picky about this... many thanks for any help.
Cas
