Hi Ben Are you trying to do this in software? We don't calculate phase directly on the ROACH (i.e compute the arctangent of the imaginary over real component) and the link shows how to correct for this, post-ROACH, phase wrapping.
What we do have in our correlator(s) here in SA (and at the GMRT) is a delay correction system that removes the phase slope in correlation products caused by physical signal path length differences. We are not really 'unwrapping' phase as described in the link but rather removing the phase slope from the data. I can provide more information if this is what you want to know. Regards Andrew
