I've been having a look at usrp2/host/lib/copiers.cc and the schematics to try and work out if the units mean anything.
As far as I can tell the LTC2284 ADC is set up to use the +-1V input range and output 14bits as a twos complement.
From the datasheet/the maths I think the ADC is thus giving possible outputs of -8192 to 8191 (i.e ~ +- 2^13) where -8192 represents -1V and 8191 represents 1V-0.000122V.
Where I'm confused is with copy_u2_16sc_to_host_32fc the integers are scaled to produce floats where -32768 and 32768 (i.e using the full 16bits of the storage variable) are mapped to -1.0 and 1.0 respectively. Is there a reason for this? I haven't tried to look at what the FPGA is doing with this data yet..
Can I change this to -8192 and 8192 to map the real/imaginary magnitudes to voltages. I think by doing this I should be able to use usrp2_fft.py -S -f xx and see the same sort of values as I see on a scope?