Since proposed by Bevis et al. in 1992, GNSS Meteorology has become a very competitive field, mainly because it provides a relatively cheap, extensive network of stations, with global and 365d/24h coverage. Therefore, the method has been widely validated by comparison with other reference techniques. A majority of the comparisons show both dry or wet bias for the PWV obtained from GNSS, depending on the particular station, as a consequence of such an heterogeneous network working on a vast set of atmospheric scenarios. Thus, most of the effort has been focused on getting the best possible global information, guarantying homogeneity. Instead, the goal of a PWV monitor for Astronomy is not homogeneity, but getting the most out of a particular station. In this work we aimed a high precision and high accuracy real-time monitor supporting IR and μW astronomical observations. We have reviewed the technique and identified the external factors impacting in the final error, including local barometric pressure and temperature measurements, weighted mean temperature based on local vertical profiles, and an independent calibration through a detailed comparison with high vertical resolution radiosondes. We discuss the final error and accuracy achieved, the lower detection limit of the technique, and possible new improvements.