2 Observations and data reduction

2.1 Observations

 

 

Table 1: Summary of the observing campaigns. The number in the last column gives the number of flat-spectrum sources observed; i.e., the steep-spectrum calibrators are not included in these figures

date	array configuration	$\nu$ [GHz]	number of sources
May 4, 1989 - May 9, 1989	full array, B $\rightarrow$B/C-reconfiguration	0.33, 1.49, 4.86, 8.44, 15.0	9
Jan. 29, 1990 - Feb. 23, 1990	sub-array, 5 antennae	1.49, 4.86, 8.44, 15.0	3
Oct. 2, 1992 - Oct. 23, 1992	sub-array, 5 antennae	1.49, 4.86, 8.44	10 (+2 stars)

The dates of the observations, the array configurations, and the observing frequencies are listed in Table 1. For the first experiment we used the full VLA for about 5 days. The array was divided into two similar sub-arrays; one of them was used for observations at 0.33GHz, 1.49, and 4.86GHz, the other at 8.44 and 15.0GHz. Both other campaigns were made during ongoing D $\rightarrow$A reconfiguration times of the VLA with sub-arrays of five antennae for periods of three to four weeks. Since the information from our observations with a five-antenna subarray was insufficient for polarization calibration, we could obtain reliable polarization information only for the first experiment. The data at 0.33GHz turned out to be heavily affected by interference and confusion problems, and could not be meaningfully searched for variability.

The observational procedure was as follows: Every source was observed for approximately five minutes per frequency, then we changed to another source. At regular intervals, we included non-variable steep-spectrum sources (1311+678 and 1634+628) to monitor and correct for any elevation-dependent and time-dependent effects (this procedure had successfully been used before in IDV observations at the 100m telescope, cf. Heeschen et al. 1987; Quirrenbach et al. 1989a). In addition, the steep-spectrum sources are useful to determine the measurement errors, since they do not show any genuine variations on short timescales (Heeschen et al. 1987).

2.2 Reduction of the total intensity data

The primary data are 10s outputs from the correlator for each antenna pair, separately for the four polarization combinations (RR, LL, RL, and LR), and two adjacent 50MHz IF bands. For the calibration, as a first step, erroneous 10s intervals were eliminated. Hereafter, we have not followed the usual data analysis procedures for VLA observations, since amplitude self-calibration would level out the variations we have been looking for. Since for sources dominated by a strong point-like core - as are most of the sources observed here - the phase is zero by definition, we have performed phase self-calibration for every source first. Subsequently, a (one-day) mean amplitude gain factor has been derived using non-variable sources like 1311+678, which were linked to an absolute flux density scale (Baars et al. 1977; Ott et al. 1994) by frequent observations of the primary calibrators 3C286 and 3C48. The choice of a constant amplitude gain factor for each one-day block of data has the advantage that systematic amplitude changes are not leveled out and can be corrected more precisely later.

During the first few hours of the May 1989 observing run, we observed the field around each source with the full array in B configuration. Maps of the sources were obtained from these data with standard VLA data reduction procedures, including full self-calibration and Högbom cleaning. The clean components corresponding to extended emission or confusing sources (i.e., all clean components but those at the origin of the image plane) were Fourier transformed back into the uv plane, and subtracted from the measured visibilities. (We assume that the flux density in the outer components does not show any significant variations.) This procedure isolates the pointlike cores of the sources. Thus, we have been able to avoid systematic artificial variations due to extended structure without losing any data (as would happen by simply discarding data outside a limited uv range). Extended structure has to be taken into account for 0716+714, 0836+710, 0954+658, 1642+690, and 1928+738 at $\lambda = 6$cm, and for 0716+714, 1642+690, 1928+738, and 2007+777 at $\lambda = 20$cm. Corrections for confusing sources within the field-of-view were made for 0917+624 and 0954+658.

After this correction and a second pass of editing spurious sections of the data, each five-minute scan was incoherently averaged over time, baselines, polarization (RR and LL), and IF bands. Due to the point-like structure of the sources (after our structure correction where necessary), the mean correlated flux density is equal to the core flux density. The flux density in the extended structure was now added back to the data, to obtain the total flux density of the source (and enable comparison to flux densities measured e.g. at the 100m telescope). Eventually, we removed spurious elevation-dependent and time-dependent effects in the light curves, using correction functions derived from observations of the non-variable sources 0836+710, 1311+678, and 1634+628.

The errors are composed of the statistical errors from the averaging and a contribution from the residual systematic fluctuations. The level of these uncorrected residuals was determined from the apparent variations of the calibrator source 3C286 and of the steep-spectrum sources. Typical standard deviations were found to be in the range of 0.5% to 1.0% (depending on the wavelength) of the mean value, with no significant difference between the individual non-variable sources. Only for the 2cm data was the standard deviation significantly higher, about 1.3% to 2%.

 

 

Table 2: Observations in May 1989, total intensity

$\lambda$ = 20cm TOT T = 5 d m0 = 0.55%
Source	N	I[Jy]	m[%]	$\chi^2_{\rm r}$	Y[%]	type
0716+714	38	0.87	1.33	5.01	3.63	II
0836+710	75	3.81	0.51	0.85		0
0917+624	101	1.13	4.27	44.26	12.70	II
0954+658	112	0.99	1.66	8.29	4.71	II
1642+690	39	1.14	1.06	2.96	2.73	II
1749+701	38	0.78	0.62	1.16		0
1803+784	38	2.13	1.80	11.80	5.13	I
1928+738	38	3.42	0.60	1.08		0
2007+777	36	1.19	1.13	3.87	2.95	II

$\lambda$ = 6cm TOT T = 5 d m0 = 0.50%
Source	N	I[Jy]	m[%]	$\chi^2_{\rm r}$	Y[%]	type
0716+714	39	0.62	3.52	54.12	10.45	I
0836+710	79	2.11	0.34	0.50		0
0917+624	98	1.51	4.38	85.09	13.06	II
0954+658	115	1.26	0.93	3.73	2.34	II
1642+690	39	1.20	0.94	3.68	2.38	II
1749+701	37	0.65	1.28	7.37	3.55	II
1803+784	38	3.48	0.58	1.56		0
1928+738	37	3.65	0.55	1.37		0
2007+777	39	2.16	0.81	2.95	1.92	I

$\lambda$ = 3.6cm TOT T = 5 d m0 = 0.70%
Source	N	I[Jy]	m[%]	$\chi^2_{\rm r}$	Y[%]	type
0716+714	37	0.69	2.83	15.56	8.24	II
0836+710	75	1.55	0.60	0.70		0
0917+624	96	1.57	2.76	6.91	8.00	II
0954+658	112	1.22	1.08	2.23	2.47	II
1642+690	38	1.17	0.96	1.78		0
1749+701	38	0.52	2.06	1.87		0
1803+784	37	3.68	0.63	0.74		0
1928+738	38	3.71	0.63	0.76		0
2007+777	36	2.35	0.53	0.55		0

$\lambda$ = 2cm TOT T = 5 d m0 = 1.30%
Source	N	I[Jy]	m[%]	$\chi^2_{\rm r}$	Y[%]	type
0716+714	39	0.97	2.10	2.58	4.93	II
0836+710	77	1.37	0.88	0.45		0
0917+624	96	1.51	2.08	2.49	4.87	II
0954+658	111	1.15	1.72	1.70	3.36	II
1642+690	38	1.13	1.38	1.09		0
1749+701	38	0.47	1.56	1.43		0
1803+784	38	3.56	1.46	1.25		0
1928+738	37	3.48	1.51	1.32		0
2007+777	39	2.30	1.40	1.14		0

 

 

Table 3: Observations in May 1989; polarized intensity and polarization angle

$\lambda$ = 20cm POL T = 5 d $m_{{\rm P},0} = 3.00$% $\sigma_{\chi,0} = 1.20$$^\circ$
Source	N	P[Jy]	$m_{\rm P}$[%]	$\chi^2_{\rm r}$	$Y_{\rm P}$[%]	$\chi$[$^\circ$]	$m_\chi$[$^\circ$]	$\chi^2_{\rm r}$	$Y_\chi$[$^\circ$]
0716+714	35	0.020	8.89	2.12	25.10	-22.36	7.53	1.59
0836+710	74	0.249	1.17	0.13		82.05	0.94	0.47
0917+624	89	0.028	10.04	5.72	28.75	35.82	2.60	1.95	6.92
0954+658	110	0.036	11.63	9.21	33.72	-31.98	4.50	8.72	13.00
1642+690	36	0.031	4.94	1.38	11.79	66.96	4.23	1.63
1749+701	36	0.013	9.48	4.80	26.98	-24.71	3.84	1.13
1803+784	25	0.101	2.76	0.70		-1.72	34.49	22.52	103.42
1928+738	37	0.062	0.82	1.25	8.68	-22.94	2.60	2.13	6.90
2007+777	33	0.027	7.89	2.70	21.88	-24.01	2.95	2.38	8.10

$\lambda$ = 6cm POL T = 5 d $m_{{\rm P},0} = 1.50$% $\sigma_{\chi,0} = 0.80$$^\circ$
Source	N	P[Jy]	$m_{\rm P}$[%]	$\chi^2_{\rm r}$	$Y_{\rm P}$[%]	$\chi$[$^\circ$]	$m_\chi$[$^\circ$]	$\chi^2_{\rm r}$	$Y_\chi$[$^\circ$]
0716+714	39	0.031	10.92	27.38	32.45	-6.15	1.77	2.44	4.74
0836+710	73	0.191	1.24	0.64		103.55	0.67	0.69
0917+624	100	0.016	27.51	101.2	82.40	-4.39	12.00	57.64	35.91
0954+658	113	0.121	2.06	1.75	4.25	-20.58	0.51	0.39
1642+690	39	0.041	8.46	17.83	24.97	-50.11	1.90	3.41	5.17
1749+701	36	0.019	6.66	9.40	19.47	-71.27	2.92	1.41
1803+784	38	0.140	4.25	7.32	11.93	83.28	1.48	1.62
1928+738	15	0.067	5.80	9.55	16.82	74.63	4.53	1.85
2007+777	36	0.047	9.96	31.28	29.53	-75.56	3.32	7.43	9.66

$\lambda$ = 3.6cm POL T = 5 d $m_{{\rm P},0} = 3.00$% $\sigma_{\chi,0} = 1.00$$^\circ$
Source	N	P[Jy]	$m_{\rm P}$[%]	$\chi^2_{\rm r}$	$Y_{\rm P}$[%]	$\chi$[$^\circ$]	$m_\chi$$^\circ$]	$\chi^2_{\rm r}$	$Y_\chi$[$^\circ$]
0716+714	35	0.043	10.21	11.33	29.29	0.23	2.09	2.46	5.51
0836+710	71	0.143	2.38	0.61		106.84	0.42	0.73
0917+624	96	0.018	15.32	14.26	45.08	-18.82	6.20	9.97	18.37
0954+658	107	0.136	2.33	0.58		-21.88	0.92	0.61
1642+690	37	0.078	4.86	2.53	11.49	-25.98	1.45	1.39
1749+701	36	0.022	5.06	2.21	12.21	-72.39	4.89	2.83	14.35
1803+784	22	0.146	2.90	0.85		-80.58	6.68	9.47	19.83
1928+738	0
2007+777	34	0.108	4.98	2.50	11.94	-71.09	1.87	2.68	4.74

$\lambda$ = 2cm POL T = 5 d $m_{{\rm P},0} = 3.50$% $\sigma_{\chi,0} = 1.00$$^\circ$
Source	N	P[Jy]	$m_{\rm P}$[%]	$\chi^2_{\rm r}$	$Y_{\rm P}$[%]	$\chi$[$^\circ$]	$m_\chi$[$^\circ$]	$\chi^2_{\rm r}$	$Y_\chi$[$^\circ$]
0716+714	37	0.066	10.34	6.91	29.20	1.42	2.31	2.03	6.25
0836+710	72	0.083	4.87	1.30		111.34	1.27	0.59
0917+624	95	0.026	13.41	5.64	38.83	-18.83	5.91	2.56	17.47
0954+658	106	0.130	3.91	0.95		-20.09	1.15	0.72
1642+690	37	0.103	6.00	2.68	14.62	-18.82	2.02	1.76	1.71
1749+701	0
1803+784	24	0.113	5.81	2.28	13.92	-70.97	9.72	6.17	29.01
1928+738	32	0.128	9.35	4.73	26.00	65.40	1.68	1.37
2007+777	36	0.163	6.19	2.36	15.32	-77.10	1.42	0.74

2.3 Reduction of the polarization data

For the polarization data (experiment in May 1989) we performed phase and amplitude calibration as described above. The Stokes parameters Q and Uwere derived from the cross correlation between the LHC and RHC polarizations, with gain factors carried over from the analysis of the total intensity data. The circular polarization of these sources is negligible, cf. Weiler & de Pater (1983). Q and U were then converted into polarized flux density Pand polarization position angle $\chi$. The source 0836+710 was used to determine the instrumental polarization, assuming that its polarized intensity and polarization angle are constant in time. In this case, the polarization vector $\vec{P} = (Q,U)$ should describe a circle in the Q-U-plane around the origin due to the parallactic rotation. The shift from the origin to the actual center gives the vector of the instrumental polarization which was subtracted from the measured polarization vector for all sources. A Faraday rotation correction was applied to the data at $\lambda = 20$cm. Since the global phase difference between LHC and RHC is unknown, all polarization angles $\chi$ may be rotated by a constant value. This was calibrated with the main calibrator 3C286 for which a polarization position angle of 33$^\circ$ was assumed at all wavelengths. Eventually, systematic effects were corrected using the non-variable sources. The resulting errors were in the range of 3% to 5% for the polarized flux density and 3$^\circ$ to 5$^\circ$ for the polarization angle. It turned out that - in contrast to the total intensity data - a significant fraction of the polarization data was very noisy. Most of these data points have been discarded, resulting in a sparser sampling for some polarization light curves. For the sources 1749+701 (at $\lambda = 2$cm) and 1928+738 (at $\lambda = 3.6$cm), we could not get any reliable polarization data.