The timing properties of Sco X-1 along its Z track with EXOSAT

We present a systematic, homogeneous analysis of all the EXOSAT ME, high time resolution data on Sco X-1. We investigate, for the first time, all power spectral properties of the < 100 Hz quasi-periodic oscillations (QPO) and noise of Sco X-1 as a function of position on the Z-shaped track traced out in the X-ray colour-colour diagram. For this purpose, we introduce a new generally applicable method for parametrizing the position of a source on its track in a colour-colour or hardness-intensity diagram. Generally, the properties of Sco X-1 vary smoothly as a function of position along the Z track. However, some variability parameters change abruptly at either of the vertices of the Z track, indicating that the branches of the Z track represent distinct source states not only in spectral state but also in rapid variability characteristics. All variability components are found to have energy spectra harder than the average flux. We show that the very low-frequency noise (VLFN) is consistent with being solely the result of motion along the Z track. The power spectra of the X-ray intensity as well as source position along the Z track extend, unbroken, to time-scales of nearly 1 day. We study the high-frequency noise (HFN) component for the first time in sufficient detail to show that there are changes in the HFN cut-off frequency with position on the Z track. It changes abruptly from ∼75 to ∼35 Hz at the normal/flaring branch vertex. The HFN is found to extend out to about 300 Hz. The QPO show a remarkably rapid change in frequency at or just before the normal-branch/flaring-branch (NB/FB) vertex. This transition happens within 1.5 per cent of the entire extent of the Z track. The QPO themselves are visible for 17 per cent of the Z. We find a new type of behaviour near the NB/FB vertex, i.e., rapid excursions from the NB into the FB and back again taking only a few minutes. We find several indications that position on the Z track is not the only parameter governing the behaviour of Sco X-1. The most dramatic examples of this are two brief episodes where the QPO frequency changed rapidly without the usual changes in colours and intensity that accompany a change of source state. In one case the frequency rapidly rose from 6 to 16 Hz, and then returned to 6 Hz, while Sco X-1 apparently remained unmoved on the normal branch in the colour-colour diagram. In the second case the QPO frequency changed from 16 to 7 Hz without the usual indications of a passage through the normal/flaring branch vertex (simultaneous dips in the count rate and colours as the frequency changes through 8 Hz). Thus it seems that deviations from the usual one-to-one correspondence between QPO behaviour and spectral state, perhaps caused QPO frequency mode switching, do occasionally occur.