a).From the integer signal processing point of view it’s better to scale all signals to 1 (e.g. –32768 to 32767 means –1 to +1). So is better to use left justified – you’ll receive data in a unitary system.

b).Let’s implement some routines for processing some signals on a uC that might have a 10-bit ADC. If you want more resolution/precision, will you review all sources to support 12, 16 bits?