Jess, Laura, Paul
We have been following up on the periodic 60Hz EY glitches which have been seen at LHO since at least June (alog 18936). These glitches are witnessed by various magnetometers and couple into DARM.
It seems that the glitches indicate the switch-on of some electronic device. Spectrograms of H1:PEM-EY_MAG_EBAY_SUSRACK_QUAD_SUM_DQ show lines at several frequencies (30, 42, 60, 80, 120 Hz) that start when the glitch occurs, then stop with a rather softer glitch about 27 minutes later. Attached are spectrograms from yesterday and last Thursday. There doesn't appear to be anything correlated in the HVE or H0:FMC channels.
We recently investigated something similar at LLO (alog 22549), which we managed to trace to an unmonitored air-conditioning unit outside the EY station. The behavior at LLO is somewhat different in that the glitch is much louder and the subsequent noise (in both PEM channels and DARM) is quieter, as well as the fact that it stays on longer each time. However, because the turn-on and turn-off are so predictable it should be possible for someone to go down to EY and listen for anything turning on and off at the expected time.
Trying to understand better the 60Hz problem. I looked at 16 hours of H1:PEM_EY_MAG_EBAY_SUSRACK_QUAD data on friday (from 2015-11-13 00:00:00 to 16:00:00) when the detector was mostly unlocked to see if the 60Hz bursts were there while the detector is unlock. As seen on the attached time plot ('60Hz_Burst_locked_and_unlocked.png'), yes they are there. This is good because that means we can look for the source while the detector is unlocked during maintenance tomorrow. In this plot I also show that the first 6 hours the detector was locked and the rest had the detector unlocked.
Each burst begins with a spike that last about 0.22 seconds (see attached figure 'Zoom-spikes_Mag_VEA_and_EBAY.png'). Notice that the spike looks different on the magnetometer in VEA and the one in EBAY_SEIRACK. While the former has a frequency of about 60.24 Hz, the later has a frequency of 121.65Hz.
Each spike is spaced between 88.5 and 90.5 minutes from each other, however in the past this was reported to be spaced by 75 minutes. After each spike there is an excess signal that last about 23 minutes, I looked at these segments of 23 minutes during the time when the detector was locked (blue arrow on the time figure '60Hz_Burst_locked_and_unlocked.png'), when the detector was unlock (green arrow on the time figure '60Hz_Burst_locked_and_unlocked.png') and to compare also when there was no burst (red arrow on the time figure '60Hz_Burst_locked_and_unlocked.png'). And then using the same color coding I plotted the spectrum of each segments around the 60Hz and its 2nd and 3rd harmonic (as attached in figures labelled as 'Spectrum_left_sideband_at_...png'). Interestingly these 60Hz harmonics show a one sided sideband at about 1Hz from the 60Hz carrier (this is particularly intense on the 2nd harmonic).
Then I looked at a Microphone (MIC_VEA_MINUSY) and an accelerometer (ACC_VEA_FLOOR) and also plotted spectrum of the same 3 time segments around the 60Hz and 2nd harmonic (figures 'Spectrum_MIC_VEA_...png' and 'Spectrum_ACC_VEA_...png' ) and although we do not see the sideband however we see a peak very near to where the sideband is at the 60Hz fundamental. This peak exist also when the burst is not there, so most probably is unrelated, but worth noticing.
Tomorrow morning we will go to EndY with portable magnetometers to see if we can notice anything. And to predict what time the spikes should occur I looked at the most recent data I have from 2015-11-16 21:15:49 to 2015-11-17 05:15:50 shown on the attached plot 'Latest_60Hz_bursts.png', we notice that the spacing is 85minutes, therefore we should spects spikes at about:
UTC LHO - Local
2015-11-17 16:02:00 8:02 am
2015-11-17 17:27:00 9:27 am
2015-11-17 18:52:00 10:52 am
2015-11-17 20:17:00 12:17 pm
2015-11-17 21:42:00 1:42 pm
