Reports 1-1 of 1 Clear search Modify search
Detector Characterisation (Broadband noise)
ruggi - 17:12 Tuesday 26 November 2019 (47797) Print this report

A decrease of OgCAVITES is often associated to a decrease of sensitivity. Starting from thurday 21 in the evening and in the following hours, such a situation was quite clear. In fig 1 we can see the Range_BNS going down together with OgCAVITIES, while other possible sources of problems where more stable (OgPR, microseism). The difference visible in the sensitivity at the beginning and at the end of that period was relevant below 100 Hz (fig. 2), not the typical situation of a 'flat noise' growing up. This is something already observed during O3a, and reported in the logbook at least twice (but I'm not able to find the entries). I tried to fit the difference, and I found something going as 1/f^1.3 (fig 3). Looking at band-rms of hrec in some clean region below 100 Hz, one can see that there was a gradual worsening for each of them (fig 4).

Some statistics over O3b data have been done, taking into account only locks in stable conditions (at least 1 hours and half after the beginning, excluding also the glitches).

Fig 5: Range_BNS vs OgCAVITIES. Low OgCAVITIES often corresponds to low Range

FIg 6: OgCAVITIES vs NI temperature. The correlation is quite good. As already known, around 19.57 C there is a good working point, corresponding to low finesse of North Arm.

Fig 7: OgCAVITIES vs WI temperature. Good correlation even there. The good working point 19.88 C is confirmed (corresponding to high finesse of West Arm).

Fig 8: OgCAVITIES vs difference of Arm Finesse (roughly computed, to be checked).

Images attached to this report
Comments to this report:
mwas - 8:24 Wednesday 27 November 2019 (47803) Print this report

In this logbook entry, there is an example of additional noise at then end of May and beginning of June. The additional noise had also a 1/f shape. It wasn't fitted using a numerical code, just by looking at the spectrum, so it may well be better approximated by 1/f^1.3. It would be good to do the fit for the May/June data, to see if the slope is the same as what is observed right now.

ruggi - 12:57 Monday 02 December 2019 (47834) Print this report

My evaluation of the slope -1.3 was just by eye too.

Here an attempt to estimate more precisely the contribution of a 1/f^1.x noise for different values of OgCAVITIES is done. Form O3b data, some segments has been selected according to the following criteria:

1) OgCAVITIES inside a certain interval;

2) low microseism (ENV_CEB_SEIS_N_50Hz_rms_0.1_1Hz<2e6);

3) low wind (ENV_METEO_WIND_SPD<25);

4) Range_BNS above a certain value (Hrec_Range_BNS>BNSmax - 2, where BNSmax is the larges value in the data that meet the criteria 1), 2), 3);

5) segments lasting at least 40 s.

Four intervals of OgCAVITIES have been selected. In fig 1 the mean value of Range_BNS for each segment is shown. OgCAVITIES has a systematic impact on the range, as already shown in a different way.

Fig 2: the spectral density, averaged over all the segments, for each OgC interval. The excess noise starts to be evident below 90 Hz.

Fig 3: the quadratic difference between the worst case and the best case. A certain shape can be recognized. I still do not know how to compute correctly that slope, but I see that -1.5 looks not bad.

More than the slope of the noise, we should try to understand if that noise is still present in the optimal condition. In my opinion, the answer is yes. In fig 4 we can see what happens if we subtract a certain amount of that noise from the curves in the best and worst case. This is not a demonstration. only a consistency check of the hypothesis.

Images attached to this comment
Search Help