The slope of phase against frequency is removed using the task FRING. It can be used to solve for a single value of delay for each antenna, polarization and IF using a small amount of data on a bright, preferably unresolved source - 2 minutes in this case. Calibrate the data prior to running the task with the appropriate CL table (DOCALIB 2). As we are only calculating a delay correction set DPARM(8)=1. Set DPARM(4) to the correlator integration time.
>TIMERANG 1 0 2 0 1 0 4 0
Write the solutions to a new CL table using CLCAL, for all sources, not just the calibration source, with INTERPOL='2PT'. If POSSM is now run as before, but with this new calibration table applied, the phases should be approximately flat across the band and the phase offsets between the IFs should be removed. Check this for your entire experiment. This is illustrated for the case of N02L1 in Fig. 5 which should be compared to Fig. 2.
An interesting application of this step is to perform a so-called ``combined IF'' solution when fringe fitting the entire dataset. As the delay differences between IFs are now removed, it is possible to solve for a single value of delay for all IFs. The advantages are increased signal to noise and, if phase referencing, consistent interpolation across target scans in each IF.