2 Observations and data reduction

The observations were performed in 1997 in four sessions, starting each day at 20:00 UT and ending at 9:40 UT of the following day, from Oct. 31 to Nov. 4.

The aim of these observations was to obtain simultaneously radio spectra at five frequencies and high resolution maps of AR Lac at 4 cm and 2 cm. The spectra were obtained with the VLA in D configuration at 5 GHz (6 cm, C band), 8.4 GHz (4 cm, X band), 15 GHz (2 cm, U band), 22 GHz (1.3 cm, K band) and 43 GHz (0.7 cm, Q band), while the high resolution VLBA observations were performed at 8.4 and 15 GHz. In order to reach the best compromise between high sensitivity for the VLBA array and the best temporal coverage for the radio spectrum, we split the VLA into two subarrays, one observing in standard interferometric mode and alternating 3 frequencies, the other in phased array mode as elements of the VLBA.

2.1 VLA data

The first VLA subarray, which consisted of the 13 telescopes equipped with 43 GHz receivers, was used to observe alternately at 43, 22 and 5 GHz, with two independent 50 MHz wide contiguous bands, in Right and Left circular polarization.

As a phase calibrator we used BL Lac, while as a primary flux calibrator we used 3C 286, which was observed at the beginning of each session at all the 3 frequencies.

To minimize the effects of the atmosphere on the phase stability at higher frequencies we observed the phase calibrator more frequently at Q and K bands than at the other frequencies. Other details of the observational strategy are given in Table 2.

Calibration and data editing were performed using the standard procedures of the AIPS package.

Since we intend to study only our radio observations taken simultaneously with the X-ray observations, which started on 1997 Nov. 2, we restrict our analysis to the last two sessions, i.e. Nov. 2-3 and Nov. 3-4. On the first two days the source was found in an active period, with flux densities reaching about 40 mJy at 6 cm. This indicates an intrinsic high variability of the radio emission, which will be analyzed in a following paper.

2.2 VLBA data

The second subarray, consisting of all the remaining 14 antennas, was used to observe in phased-array mode at X and U bands as an element of the VLBA array. The presence of half the phased VLA, corresponding to a 90 m telescope, increased the sensitivity of the VLBA.

The data from the VLA phased-array can be also analyzed as a standard interferometer, therefore total fluxes were computed for all the observed sources. The VLBA observations were performed alternating between X and U bands, with a typical scan lasting 45 min. For each observing frequency, we observed in dual polarization mode, with a total bandwidth of 64 MHz. Since AR Lac has generally a low flux density (from few to some tens of mJy), we used the phase-reference technique (Beasley & Conway 1995), consisting of the rapid switching between a strong calibrator, close to the target source, and the source itself. In this way, in the successive analysis, phase calibration of the calibrator can be applied to the target source. We used BL Lac (3.6 degrees apart from AR Lac) as the reference source. The data tapes were correlated at the VLBA correlator at the Array Operation Center in Socorro (New Mexico).

For VLA data, calibration and data editing were performed using the standard procedures of the AIPS package. The flux calibration was performed by using the measured system temperature and gain curves for VLBA telescopes and from the measurements of the ratio $T_{\rm ant}/T_{\rm sys}$ made on the calibrator BL Lac for the phased-array, once the flux density of this source was determined from the VLA data. Delay and delay rate were determined for BL Lac, which was then self-calibrated. The final map of the self-calibration was used as a model for the final global fringe fitting, whose solutions were applied to our target. A preliminary map of AR Lac at 8.4 GHz was made to check its position, and it was found about 14 milliarcsec (mas) away from the phase tracking center. This was due to the fact that the coordinates used at the correlator were the heliocentric ones, and the displacement of the source was due to the annual parallax. The visibilities of AR Lac were then phase-rotated in order to bring it to the phase center. The operation makes it possible to time average the data over a relatively long time (tens of minutes) without loss of signal in the longest baselines, allowing us to analyze the amplitude of the visibility as a function of the baseline length. If the source is not at the phase center, a rapid change of phase would occur at the longest baseline.

Figure 1: Top panel: VLA flux curves of AR Lac binned over 20 min. Dot-dashed vertical lines indicate the times of the spectra labelled by the letters on the top and shown in Fig. 2. Lower panel: the contemporaneous SAX light curves (from Rodonò et al. 1999).

The source was successfully detected at 8.4 GHz, but not at 15 GHz, probably because of a poor phase stability at this frequency, which we could not correct with the phase referencing technique.

Figure 2: AR Lac VLA spectra obtained at times indicated by dashed vertical lines in Fig. 1.