CAZIMUTH bug in WSW IMBFITS?

The horizontalTrue system was introduced to properly measure offsets (distances) on the sky, as the difference in azimuth coordinates is not the actual distance along azimuth. On the other hand, absolute azimuth-elevation coordinates are what they are, with no particular issue.

However, it appears that the CAZIMUTH (commanded azimuth) found in the IMBFITS files is incorrect with respect to this definition. Fig.  shows an example where CAZIMUTH is $105^{\circ}$ and AZIMUTH is $154^{\circ}$ for a $47^{\circ}$ elevation at start of the scan, noting that $154 \times cos(47) \simeq 105$. This is even more surprising at this feature is only (and always) observed in wobbler-switch IMBFITS files, inconsistently with the other observing modes.

Note that MRTCAL uses CAZIMUTH and CELEVATIO 1) to give a typical (az,el) position of each IMB-FITS at indexing time, and 2) to give a (r%head%gen%az,r%head%gen%el) position for each CLASS spectrum it creates. CLASS relies on commanded values, not the actual from antenna-fast tables.

Figure: Commanded (red) and Encoder (black) azimuths and elevations as found in the antenna slow and fast traces respectivly in the named IMBFITS file.