It would be nice if the DSPSR library correctly computed the frequency response function used for phase-coherent dispersion removal by correcting the Doppler shift of radio frequencies and time delays owing to the motion of the observatory with respect to the Solar System barycentre.
Some experimentation started on the feature-requests-30-barycentric-dispersive-delay-correction branch. Note that the current calculations are incorrect; they correctly convert topocentric radio frequencies to barycentric radio frequencies, but do not convert the barycentric delay back to a topocentric delay.