A brief description of ROG DATA


The Detectors

The ROG groups operates two Gravitational Wave (GW) detectors : Explorer (at CERN-Geneva) and Nautilus (at LNF-Frascati).
They both consist of a 3 m long Aluminum bar, suspended in vacuum in a cryogenic environment. The amplitude of vibration of the first longitudinal elastic mode of the cylinder, resonating around 900 Hz, is the quantity monitored and analyzed in search of an event generated by GW.
The bar vibration is converted into an electric signal by means of a resonant transducer, constituting a capacitor with a very small gap (12 &mu m or less). The output of the transducer is first amplified by a d.c. SQUID device, then by other more standard electronics and finally sampled by an ADC at 5 kHz sampling frequency.
In order to reduce the thermal noise, both antennas are cooled down to about 3 K. A suitable suspension and isolation system is used to suppress the transmission of seismic noise from the ground to the antennas.
Typical spectral sensitivity curves of the antennas can be seen here.

Analysis of the Data

The more standard type of analysis performed is aimed to detect short (few milliseconds) impulsive excitation. The continuous data stream produced by the sampling ADC is acquired by a VME real-time CPU, transferred to a computer and stored on disks. The analysis code first performs a reduction of the 5 kHz data by selecting a frequency band, about 160 Hz wide, around the sensitive region in the FFT of the data. The spectrum is then shifted down, so as to start from zero, and finally an inverse FFT is performed, thus obtaining a time sequence at 3.2 ms sampling time. To these data (called RAW data) is then applied a filter optimally matched to the response to a delta-like excitation of the bar, thus producing the so-called filtered data (the FIL files we make available).

Searching for Events

The amplitude of the data in the filtered stream has a nearly Gaussian distribution. Searching for candidate events is done looking for values of the amplitude high with respect to the sigma of the data themselves. A threshold is set, usually in terms of critical ratio defined as CR = (|x|- avg(|x|)/&sigma(|x|). An event is thus defined by the time and amplitude of the local maximum is the data over threshold. A dead time needs to be defined, as a minimum separation between adjacent events (usually 1 second).
Starting from 2005, the search for events is performed on the filtered data streams at 5 kHz sampling time. This allows a better definition both of the amplitude and of the time of the events.