Astron. Astrophys. 347, 203-211 (1999)

3. ROSAT HRI observations and results

3.1. YY Draconis

Here we report details of the longest X-ray observation yet of YY Dra. It was made with the ROSAT High Resolution Imager (Zombeck et al. 1995) and comprises 56.6 ksec on source between 1996 May 22 23:54 UT and 1996 May 25 06:27 UT. A lightcurve at 10 s time resolution was constructed by using the Starlink Asterix software (Allen & Vallance 1995) to optimally extract data from a region 0.68 arc min in radius, centred on the source. Background subtraction was carried out using the data from a nearby region of sky, and the resulting lightcurve is shown in Fig. 1 at a lower time resolution. The lightcurve is broken-up by the 90 minute satellite orbit and has a mean count rate of 0.40 c s^-1.

Fig. 1. The ROSAT HRI lightcurve of YY Dra at a time resolution of 100 s. There are no data in the time intervals between those illustrated in each panel.

To analyse the lightcurve we used the 1-dimensional CLEAN algorithm in the implementation of H.J. Lehto. This is particularly suited to time series which are irregularly sampled and in which multiple periodicities may be present. Fig. 2 shows the resulting raw power spectrum, window function, and CLEAN ed power spectrum. The only significant signal detected is at a frequency corresponding to a period of () s, and may be identified with the first harmonic of the white dwarf spin frequency described earlier. Other spikes in the power spectrum, at a level of  c² s^-2 or lower, do not correspond to any known system period, or combination of periods, and are presumed to be due to noise. The power at the first harmonic of the spin frequency is  c² s^-2 which corresponds to an amplitude of 0.040 c s^-1. (Nb. The amplitude is equal to twice the square root of the CLEAN ed power.) The modulation therefore has a fractional amplitude of 10%.

Fig. 2. The power spectrum of the ROSAT HRI lightcurve of YY Dra. The upper panel shows the raw power spectrum with the window function inset. The lower panel shows the CLEAN ed power spectrum. Tick marks show the expected locations of the orbital and spin frequencies, along with some of their sidebands and harmonics.

We note that no significant power is detected at either the orbital period or the beat period of the system, or at any harmonics and sidebands of either of their corresponding frequencies, in these data. In particular, the two peaks with power of  c² s^-2 at frequencies of about  Hz (Fig. 2) are not coincident with any harmonics of the orbital frequency; nor is the broad dip in the lightcurve around  s (Fig. 1) believed to be related to any orbital phenomenon.

The limit on the power at the actual spin period of YY Dra (529.3 s) is  c² s^-2, corresponding to a limiting amplitude in the light curve of  c s^-1, and a limiting fractional amplitude of less than 1.5%. However, given the abundance of `noise' peaks with power of the order of  c<sup>2</sup> s<sup>-2</sup>, a more realistic estimate of the limiting fractional amplitude is around 3.5%. The data folded at the white dwarf spin period are shown in Fig. 3. From this, it is clear that the pulse profile can be described as `double-peaked' with two similar peaks per cycle separated by about 0.5 in phase. Since the power at this period is clearly very small, it is not surprising that the difference between the two peaks is negligible.

Fig. 3. The ROSAT HRI lightcurve of YY Dra folded at a period of 529.3 s. Phase zero is arbitrary, and the profile is shown repeated over two cycles.

3.2. V709 Cassiopeiae

As with YY Dra, we report details of the longest X-ray observation yet of V709 Cas, made with the ROSAT HRI. The observation comprises 43.3 ksec on source between 1998 Feb 15 14:31 UT and 1998 Feb 17 08:45 UT. Again using the Starlink Asterix software, data were optimally extracted from a region 0.65 arc min in radius, centred on the source, in order to construct a time series at 10 s resolution. Background subtraction was carried out using the data from an adjacent region of blank sky and the resulting light-curve, shown in Fig. 4 at a lower time resolution, has a mean count rate of 0.26 c s^-1.

Fig. 4. The ROSAT HRI lightcurve of V709 Cas at a time resolution of 100 s. There are no data in the time intervals between those illustrated in each panel.

As above, we used the 1-dimensional CLEAN algorithm to analyse the light-curve and Fig. 5 shows the results. We detect signals at frequencies corresponding to the previously identified spin period and its first and second harmonics. Using the accurately determined frequency of the second harmonic we calculate a more precise value for the spin period than previously measured, namely  s. The power at the fundamental frequency is  c² s^-2 which corresponds to an amplitude of 0.059 c s^-1. The modulation therefore has a fractional amplitude of 23% and the data folded at a period of 312.78 s are shown in Fig. 6.

Fig. 5. The power spectrum of the ROSAT HRI lightcurve of V709 Cas. The upper panel shows the raw power spectrum with the window function inset. The lower panel shows the CLEAN ed power spectrum. Tick marks show the expected locations of the spin and orbital frequencies (assuming an orbital period of 5.4 hr), along with some of their sidebands and harmonics.

Fig. 6. The ROSAT HRI lightcurve of V709 Cas folded at a period of 312.78 s. Phase zero is arbitrary and the profile is shown repeated over two cycles.

As with YY Dra, we note that no significant power is detected at either the orbital period (whether 5.4 hr or 4.45 hr) or the beat period of the system, or at any harmonics and sidebands of the frequencies corresponding to either period. The limit to the power at any other period is of the order of  c² s^-2, corresponding to a limiting fractional amplitude of about 2.5%.

© European Southern Observatory (ESO) 1999

Online publication: June 18, 1999
helpdesk.link@springer.de