In my previous report, I missed one further case of bad data
occuring 18 seconds prior to a band change:
For the upper pair of frequencies at C-band (5700 and 6750), both
IFs had bad data for the 17 seconds prior to the end of the scan. The
following scan was at X-band.