I'm reading from some cRIO cards that output <+/-,24,10> FXP's which is *much* wider than I need it to be. I'm reading in voltages that are always positive and less than 100V and I have a 1mV accuracy requirement.
Is it good practice to immediately add a "To Fixed-Point" primitive after the I/O node to reduce a <+/-,24,10> to a <+,17,7> (in this case) to save 7 bits per I/O channel and possibly more if I do any sort of math upon them?