3. Light curves
The background-subtracted light curves of the two flares as measured with the WFCs are presented in Figs. 1 and 2. Besides these two occasions, the source was never detected in either short 1-hr or long 24-hr WFC exposures with typical upper limits of 50 and 12 mCrab respectively. The light curves of both detections are characterized by sporadic flaring activity up to 0.2 ct cm-2 s-1, which corresponds to 0.1 Crab units, with a larger and longer flare on Sep. 10 with a peak of 0.3 Crab units. The large flare was above the detection limit for about 6 hr and occurred in the middle of a 34 hr long observation. It started on Sep. 10.0, which is 5.4 d before the giant outburst. Its duration is comparable to that of the giant flare.
The flaring activity on a time scale of hours is in line with the behavior as measured with the Proportional Counter Array (PCA) on RossiXTE (Markwardt et al. 1999b). The PCA covers SAX J1819.3-2525 bi-weekly since Feb. 5, 1999, for individual snapshot exposures of about 60 s with a sensitivity of about 1 mCrab. SAX J1819.3-2525 was first seen on Feb. 18 and subsequently showed an erratic variable behavior until immediately after the giant outburst, with fluxes ranging from 0.5 to 30 mCrab (Markwardt et al. 1999b). Since the giant outburst, no emission was detected anymore (C.B. Markwardt, priv. comm.).
The WFC data do not reveal any type I X-ray burst from SAX J1819.3-2525 in about 0.8 Ms of source coverage for the year 1999 when it was seen to be active with the PCA (Markwardt et al. 1999b), or for 3 Ms over all WFC observations since 1996.
Fig. 3 shows the evolution of the background-corrected photon count rates in various bandpasses of the three NFI, in 400 s time resolution. For an explanation of the method of data extraction, we refer to In 't Zand et al. (1999b). The light curves show slow variability on a time scale of a few hours with an amplitude of about 50%. The average flux level is 12 mCrab (2-10 keV) which, relative to the flares, can be regarded as calm emission. However, we note that this cannot be regarded as quiescent emission because the PCA has seen flux levels from this source at least one order of magnitude smaller (Markwardt et al. 1999b).
A power density spectrum of the MECS photon count rate, generated with a timing resolution of 0.02 s and averaged over 256 s time intervals, reveals no measurable narrow features nor broad-band noise. The upper limit on the variability, integrated between 0.01 and 10 Hz with power laws with indices of -0.5, -1.0 and -2.0, are 18, 8 and 3.5% fractional rms respectively (90% confidence). The upper limit on the pulsed amplitude is 1.3% fractional rms (95% confidence) in 0 to 25 Hz.
© European Southern Observatory (ESO) 2000
Online publication: June 5, 2000