This page lists some of the basic hardware and operations issues that affect MARVELS data. In DR12 there was an effort to remove or fix these problems. However, there are still many bad radial velocity data points that have made it through the DR12 pipeline. In DR11, it is necessary to detect and remove any radial velocity points that are suffering from these problems.
Sometimes, the optical fibers that are plugged in at the telescope to observe the desired stars do not fit as snugly as needed, and occasionally, they will fall out. This is called a dropped fiber, and leads to no light being observed for that star as long as the dropped fiber is not plugged back in. At worst, a dropped fiber might go undetected for an entire bright-time observing period- a couple of weeks- but at best, a dropped fiber might be fixed by the next day. A dropped fiber may be detected by noting whether there are some observations with peculiarly low photon counts.
It is also possible for the optical fibers to break, which also leads to little or no light being observed for that location. While these "dead" fibers can be repaired, that typically happens only at the end of a bright-time observing period, so several observations in a row through that dead fiber may be junk. A dead fiber may be detected by noting whether there are some observations with peculiarly low photon counts.
Sometimes, the optical fibers get plugged into the wrong holes at the telescope. This is called a misplug. Most often, this takes the form of a swap- two fibers get switched with each other (however, larger-scale instances can also occur, if, for example, an entire bundle of fibers gets mounted incorrectly inside of a fiber cartridge for a couple of months). This leads to the wrong star being observed in a particular place on the CCD. Unfortunately, because the instrument characteristics vary depending on the position of the final spectrum on the CCD, such misplaced data may be difficult or impossible to calibrate with the same accuracy as normal data. Misplugs are often identifiable because the photon count for a misplugged star will be much higher than normal. Often, the pattern of spectral lines on a misplug is also quite different, since a different star than usual was observed in that location. Misplugs can also be identified in the final RV curves, because a misplugged star usually has a very different space velocity than the star which was supposed to be observed in that location.