Yesterday I ran a cable calibration prior to a test and got expected values for it. I re-ran the same calibration today and now have a 13dB loss across the entire test band (i.e. a flat response 13dB below what's expected.). I've stripped the setup down to running a known to be good RF cable between a signal generator and the input of the SA and see the same 13dB loss as was measured in the test setup. I've swapped out the signal generator, and tried multiple cables with the same results. I've also done a reset to factory settings and reboot of the SA to no effect.
Everything seems to point to the SA having a fault, but before I send this in for service, I'd like to ask for the community's input as this seems like an odd way to fail, and I want to eliminate the possibility of user error in the instrument setup.