2. Observations and data reduction
The VLBI observations used an array consisting of the VLBA 1 and Effelsberg 2, Medicina and Noto 3 at 22 GHz, while the stand-alone VLBA was used at 43 GHz. Five sets of observations were made from December 1992 to January 1993 at roughly 10 day intervals. Data were recorded on Mark IIIA compatible terminals, in Mode E (14 MHz bandwidth). Typically, each scan had lasted 13 minutes. The data were correlated at the Array Operations Centre in Socorro, New Mexico. The correlator output was calibrated in amplitude and phase using 4 and imaged using DIFMAP 5 (Shepherd et al. 1995).
The amplitude calibration of VLBI observations at 22 GHz and 43 GHz, is mainly limited by the lack of point source calibrators. System Temperature measurements were made at the end of each scan and the gain curve of each telescope was used to correct the variation of effective antenna gain with antenna elevation for all telescopes. The calibration accuracy differs between observing sessions, owing to different weather conditions. (Millimetric observations are sensitive to the water vapor content of the atmosphere.)
Images were made for each `calibrator' source scheduled for a few scans during the observing program, namely 0804+499 and 1611+343 at 22 GHz, OJ 287 and BL Lac at 43 GHz. From an inspection of the visibility plots, it appears that these sources were barely resolved with the array and are therefore reasonable calibrators. The poor uv -plane coverage (3700 and 5000 visibilities on average at 22 GHz and43 GHz respectively) and the rms noise ( 14 mJy) did not allow these to be mapped properly. Much of the flux from their jet structure (see Kellermann et al. 1998) is clearly lost. We were able to image the central brightest component only and most of the emission from the more extended structure is missed.
The self-calibration procedure, which uses closure amplitudes to determine telescope amplitude corrections, gave calibration factors that are within 10% of unity. Total power measurements, contemporary with the VLBI observations, were not available for the calibrators. These sources are known to be highly variable with time at high frequencies. Light curves at 22 GHz for 1611+343 and OJ 287 can be found in Tornikoski et al. (1994). The ratio between our extrapolated `zero baseline' flux density and the total power flux density is 0.75 for both 1611+343 and OJ 287. Tornikoski et al. (1994) also report monitoring observations for BL Lac at 37 GHz. Here, the ratio between the two flux densities is 0.5. However, BL Lac has a more extended structure than 1611+343 and OJ 287 and we were unable to image this successfully.
The total correlated flux density for 3C 273 at both frequencies is about one half of the total power measurements with single dish. However, correlated flux and total power flux follow a similar trend (see, for example von Montigny et al. 1997). More recent 22 GHz VLBA observations by Leppänen et al. (1995) do show a similar ratio. In conclusion we estimate that the systematic amplitude calibration errors for any of the data sets are .
During session four (14 January 1993), the array was so heavily affected by adverse weather conditions that only crude images could be made. Data from that session will not be used in the following discussions. Table 1 summarizes the observational data and imaging. Column 1: project name; Column 2: date of the observation; Column 3: observing frequency; Column 4: stations; Column 5: beam major axis; Column 6: beam minor axis; Column 7: major axis Position Angle; Column 8: dynamic range, i.e. the ratio between the peak flux density and the rms noise in the image, measured far from the source of emission.
Table 1. VLBI observations of 3C 273.
© European Southern Observatory (ESO) 1999
Online publication: May 21, 1999