Reports 1-1 of 1 Clear search Modify search
Detector Characterisation (Broadband noise)
ruggi - 16:15 Saturday 17 August 2019 (46691) Print this report
Missing 4 Mpc

When the range passed from 48 to 42 Mpc, one identified guilty was the worsened working point. Today, as already pointed out by Matteo, the working point moved back towards the good condition, and some improvement of the sensitivity has been observed. Here a plot showing the integrated range is added (fig 2): almost 2 Mpc has been gained between 60 and 150 Hz. I think this is mainly due to PRCL noise which is gone, but Michal can evaluate this more precisely than me.

4 Mpc are still missing. The working poiint could be not yet perfect, perhaps a check of alignment setpoint could provide the usual improvement of high frequency CMRF and DET scattered light coupling, but an improvement of 4 Mpc is not expected.

Looking at the sensitivity (fig 3), a problem at the level of high frequency floor is visible. A check of SQZ performance might be useful. If something is wrong there, maybe also some structure (40 Hz, 55Hz, 120 Hz, 160 Hz) could be the scattered light already seen when SQZ was misaligned?

Images attached to this report
Comments to this report:
mwas - 15:44 Sunday 18 August 2019 (46696) Print this report

I can confirm that the PRCL noise is mostly gone, however it is still higher than before Aug 7. There is still some contribution, especially below 50Hz, but the impact on the range is small. With offline h(t) cleaning I manage to improve the range only by 0.7Mpc.

Images attached to this comment
ruggi - 11:38 Monday 19 August 2019 (46701) Print this report

The improvement of the sensitivity did not survive so much this week end. A deterioration tendency has been observed (fig 1), and it seems not due to PRCL noise coming back (OgPR looks quite stable - fig 1). Comparing a 45.6 Mpc and a 43.4 Mpc, the main difference is below 100 Hz (fig 2), but the integrated range (fig 3) says that the worsening starts at higher frequency, and it is smoothly distributed.

An attempt to estimate the additional noise has been done, finding as best fit a 1/sqrt(f) (more precisely: 4.7e-24/(f/100)^.5). In fig 4 and fig 5 one can compare the noisy data and the clean data plus the noise (added in quadrature).

Among the parameter which uses to be correlated to the sensitivity, I found as best candidate OgCAVITIES (fig 1, but the variation seems really small).

This analysis is in agreement with some similar observation done by Michal at the time of the first sensitivity drop.

Images attached to this comment
ruggi - 13:03 Monday 19 August 2019 (46703) Print this report

The attached plot shows the statistical distribution of the RANGE for different values of OgCAVITIES (all O3 data). For each Og value, there is a maximum value of reachable range, linearly correlated to the Og value. In other words, low values of Og are correlated to some additional nose which prevents the range to overcome a certain value.

Images attached to this comment
ruggi - 0:35 Friday 23 August 2019 (46748) Print this report

Today we had again 46 Mpc (fig 1). The good level achieved 2 weeks ago is not yet back; there is still more noise between 40 and 60 Hz, at 160 Hz and in the high frequency noise floor. In fig 2 impact of those structures in terms of Mpc is more clear.

The good sensitivity of early august is not so good as the one of early april (fig 3). Fig 4 says that there is a difference of 4 Mpc well distributed from 60 Hz to high frequency, but in the recent data we gain 1.5 Mpc due to an improvement at 30-55 Hz.  In fig 5 and fig 6 the good data have been worsened adding the old flat noise (5.1e-24*(f/100)^0.25), and this seems to fill the gap very well.

Images attached to this comment
mwas - 15:59 Monday 26 August 2019 (46769) Print this report

We could check if the "flat" noise we have been chasing is at the origin of this noise increase, by redoing a comparison of the noise with the nominal dark fringe offset, and with double the power on the B1 photodiodes as was done in the second half of shift done back in June.

 

Search Help
×

Warning

×