Virgo Logbook

Environmental Monitoring (Environmental Monitoring)

Paoletti, Fiori - 14:21 Tuesday 06 November 2018 (43422)

new lines in DARM - a reminder

Comparing tonight lock with the last long ones (Oct. 27th), some differencies in DARM are:

- a "bump" around 290hz disappeared

- two new lines appeared at exactly 35hz and 63.5Hz (maybe calibration lines?)

- some "arches" appeared, with a period of about 10 minutes, between 650 and 800 Hz (see fig. 1)

- a new "triplet" appeared around 3276Hz (sidebands ± 0.7Hz) moving in frequency with time (see fig.2)

Images attached to this report

Comments to this report:

swinkels - 18:07 Tuesday 06 November 2018 (43427)

I ran my line tracker (now tentatively called the Brute-force Utilities For Finding Annoying Lines and Others) on the moving line around 670 Hz. The line is very faint, so I had to help a bit by defining the right region to find the maximum. The channel with the best correlation is a temperature sensor inside the LNFS DBOX, which flips by just a single bit. Further down the list are other channels related to the LNFS and signals from the HVAC system of the injection system. The source of the noise is likely something related to the (demodulation of the) LNFS, which is modulated by the temperature of the room. As a check, I also ran the analysis over 4 hours of data without manually defining a region, which yields very similar winning channels.

Images attached to this comment

Non-image files attached to this comment

correlate1225553204.txt

mwas - 10:49 Thursday 08 November 2018 (43465)

Figure 1, the same moving lines at 600-800Hz are visible in PRCL (with a better SNR). The most likely scenario is that the SSFS error signal gets spoiled by something in the 56MHz modulation, and then that extra noise is impressed on the laser frequency by the SSFS loop, and LSC_PRCL at high is a nice out-of-loop witness of the laser frequency noise
Figure 2, shows that these moving lines are not present in h(t), as the reconstruction is subtracting the laser frequency noise (using the PRCL error signal)
Figure 3, shows a bit more of the PRCL spectrum, the moving comb of lines has a separation of about 47Hz, it is clearly visible between 500Hz and 900Hz, and there is one more line with the same time evolution around 230Hz.
It is not a real moving comb, with the same time evolution shifted in time frequency, but the harmonics of some fundamental, as the time evolution are scaled (the line around 830Hz line has a much larger dynamic than the line around 230Hz)

Figure 4, shows that the base line is also visible in PRCL between 47 and 49 Hz.

Images attached to this comment

mwas, masserot, letendre - 13:20 Thursday 08 November 2018 (43472)

Figure 1. The wandering lines are also clearly visible in the LNFS 56MHz demodulated signal phase. After discussing with Nicolas and Alain the noise source could be the 100MHz quartz oscillator inside the LNFS DAQ box. This oscillator is free running, so it frequency will change with temperature. The oscillator is not used for any parts of the demodulation, but there might be an electrical cross-talk with the demodulation that uses the 100MHz supplied from the timing distribution system, which then shows up as a beat between the two 100MHz signals.

A simple 5 minute test of this theory would to change the temperature of the 100MHz free running oscillator (it is driven by a DAC). If all the bumps in the LNFS channels move, then it will be a smoking gun that the free running oscillator is causing the wandering lines.

Images attached to this comment