Dropping V Cs
To: Stations Observing in Geodetic VLBI Experiments
From: Ed Himwich and Brian Corey
Re: Dropping Channels Gracefully
Ideally, all stations should have a full complement of working BBCs (or VCs, both hereafter referred to as "BBCs"). However, we recognize that there may be occasions when this is not possible. For many years we have had a standing recommendation that, for experiments that require 14 BBCs, a station with a single bad BBC should put it in position #6, i.e., BBC #6 should be "dropped". If a second BBC is bad, BBC #11 should be dropped. We hoped that the bad BBCs would be repaired rapidly so that a full complement would soon be available again. The point of these recommendations, originally developed by Dave Shaffer in the 1980s, is two-fold: (1) to have the least impact on the delay resolution function, in terms of both preventing the wrong peak from being picked in fringing and losing the least amount of observation precision, and (2) to avoid having different stations, operating independently, drop different BBCs, thereby causing even more data loss between them.
Due to the aging of equipment and the difficulty of obtaining repairs, there are times now when stations may have more than two BBCs that are not working and/or they may not be being repaired in a timely fashion. Consequently, Brian, with contributions from Alessandra Bertarini, has put together recommendations for which BBCs to drop when missing up to four. We are not planing to make a recommendation for going beyond dropping four BBCs for experiments that require 14 due to the rapid reduction in performance that occurs. Please note that the recommendations apply to all experiments requiring 14 BBCs. Although the recommendations are not optimal for all frequency sequences currently in use, they are nearly so. Very little improvement would be achieved by varying the recommendations for different experiments. Having to keep track of which BBCs should be dropped in which experiments would reduce the utility of the recommendations beyond what is justified, given the small improvement that would result. If a new frequency sequence is developed that requires a different drop order, the recommendation will be revised; for the foreseeable future however this is not a concern.
For now, there are only two qualification to the experiment independence of the recommendations: (1) for the infrequent experiments that require only eight BBCs (RDVs primarily), stations should physically re-arrange the BBCs if necessary in order to have a full complement of BBCs, and (2) if a station knows that one or more channels are persistently lost to RFI or some other problem, they should try to use the bad BBCs in those positions first before applying the drop order. The second qualification has reduced importance compared to general recommendations; if applied it may lead to varying the drop order depending on the experiment. The following final paragraph of this message states the new recommendations independent of such complications.
For experiments requiring 14 BBCs, when a station does not have enough working BBCs (or VCs) to form a full set, we recommend the following order for dropping BBCS: 6, 11, 7, and finally 2. In other words: if a station has a single bad BBC, they should place it in position 6; when there are two bad BBCs, they should go in positions 6 and 11, and so on. The order of dropping BBCs is independent of the frequency sequence for experiments using 14 BBCs.