This is interesting. It looks like the low-frequency noise (DC to 50 Hz circa) is real magnetic noise emitted by the system, because it is also visible in the coil current monitoring channel.

It comes from a broadband noise, as visible in the coil voltage monitoring channel, that is shaped by the coil inductance (the power amplifier is in Voltage Mode Driving).

We have to understand whether it is due to the DAC channel, or to the Interface Electronics between the DAC and the power amplifier, or finally to the Power Amplifier itself.

During the test we noticed at some point several missing samples also 10s long (dark gray in the dataDisplay). Seems that the  problem disappeared after restarting the ENVnoise process.

Today we did a further step in testing the ENV node. We also performed a full set of magnetic sweep injections.

First we tested the ENV_MAIN node manually.

At 8:58 UTC we run a short test of the sequence (duration times set at ~60s for each action) with the ITF locked: magnetic CEB, magnetic NEB, magnetic WEB, and acoustic in CEB. The ITF unlocked at 9:10:31 UTC right when the acoustic CEB injection started. The reason is that we set a too loud volume (0.04 V instead of the 0.004 V used successfully on Dec 16). Figure 1 shows some signals that look correlated to the unlock. The F0 of several suspensions were excited.

At 11:32, right after reaching LowNoise3 the ITF unlocked because of an earthquake (Mag. 6.5 from Japan, 10:15:16 UTC).

At this point we decided to test the ENV_MAIN node in automatic mode. 

At 12:08 UTC the ITF was again in LowNoise3

After 900s (12:24 UTC) the ENVnode started automatically the sequence of injections, with the following parameters:

• CEB magnetic sweep: range 1 - 1000 Hz, duration 1800s, amplitude 9V
• NEB magnetic sweep: 1) 8-80 Hz, 300s, 1 V; 2) 40-100 Hz, 300s, 2V; 3) 70-1000 Hz, 600s, 4V (note that we reducted the amplitude because of the unlock we expeienced yesterday)
• WEB magnetic sweep: 1) 8-80 Hz, 300s, 1 V; 2) 40-100 Hz, 300s, 2V; 3) 70-1000 Hz, 600s, 4V
• CEB acoustic colored: range 8-2000 Hz, duration 300s, amplitude 0.004 V

The magnetic injections went smoothly (data to be analyzed). Instead the acoustic injection again caused an unlock at 13:37:36 UTC. Figure 2 shows some signals correlated to the unlock. The microphone signal shows a louder peak emission at the start of the injection (the reason has to be better investigated). Right in correspondence to it the BS F0 TY signal starts to drift significantly. Also other signals show a similar behaviour, like the OB, the SR, the WI, the NI (not so clear). Not the IB, the NE and the WE

Test of acoustic injection with the ENV_MAIN node

We did several attempts of injecting colored (8-2000 Hz) acoustic noise in the CEB trying to eliminate the problem of the louder spike at the start of the injection.

First we used the automated node. Il seems that the spike appears when the volume of the injected noise goes above 0.001 V (despite each time we manually issued a "reload filter" and "enable filter"). Figure 1 shows one example with the spike (volume at 0.004 V, this one caused an unlock of the ITF). Figure 2 shows one example without the spike (volume at 0.0004V. This is too low for our purposes).

Then we did some trial with the ENVnoise process (but we set off the loudspeaker to be safe). In this case we could inject up to 0.004 V without the spike. This is shown in Figure 3: the two tests done between 10:25 and 10:45 UTC are with a volume of 0.004V and there is no spike.

Test and actual injection of magnetic lines in the range 2 - 25 Hz in CEB, using the ENV_MAIN node

After some tests, at 14:15 UTC we ran the ENV_MAIN node to inject the following magnetic lines with the CEB large coil (in a sequence):

•   2.5 Hz, 3.5 Hz, 4.5 Hz, 5.6 Hz, 6.5 Hz, 7.5 Hz, 9.0 Hz, 10.6 Hz (each for 900 s)
•  12.0 Hz, 13.6 Hz, 15.0 Hz, 19.0 Hz, 22.0 Hz, 24.0 Hz (each for 600s)

The run should end at about 17:15 UTC. The gps of the single injections will be written in the file

/virgoData/NoiseInjections/MagneticInjectionsO4/output/MagneticLine_NOISE_MAG_CEB-1387894545.txt

Indeed it looks that the lines are detected in Hrec. Figure 4 shows the example of the 2.5 Hz line.

The ITF unlocked at 17:00:29 UTC. The script had almost completed. It was injecting the 22.0 Hz line since a few minutes. Figure 1.

Unfortunately the log file with the time of the injections remained empty. This has to be fixed in the injection code: the gps has to be written at the end of each single injection, not at the end of the process.

The ITF reached LowNoise3 at 17:36:59 UTC.  We (Andrea M) set the ITF mode in DQSTUDIES.

We set the ENV_MAIN node to start automatically at 6:30 UTC of tomorrow Dec. 30th. When low noise 3 is reached, and the ITF will be set in the COMMISSIONING mode, the ENV_MAIN node will wait for 900s and then will perform 1 set of magnetic lines in the NEB large coil (for a total duration of 3 hours) and then the same set in the WEB large coil (another 3 hours).

NOTE for the Operator:

Please tomorrow morning change set the ITF in COMMISSIONING mode !!!

This comment is related to the event shown in the figure1

Below the detail of the commands received by the ENVnoise Acl server . To be noted that all the commands sent by the  "AcousticInjectionCEB" are with "load and apply" each time

• To set tha noise_white amplitude
  • 2023-12-29-09h50m13-UTC>INFO...-AcNoiseChAmplCmSet> noise_white, ampl 0.01 rampTime 0.1 - loaded and request to apply received - sender "AcousticInjectionCEB"
  • 2023-12-29-09h50m13-UTC>INFO...-AcNoiseChSet> noise_white - ampl 0.01 - rampTime 0.1s - seed 0 - cnt_max 15
  • 2023-12-29-09h50m13-UTC>INFO...-AcEngStateSet> state 3(4) - nload 1/40(0) - nLoadFake 39 - cntLoad 86
  • 2023-12-29-09h50m13-UTC>INFO...-AcCompGpsLoadSet> GPS1387878632/0 - stsGps 1
  • 2023-12-29-09h50m13-UTC>INFO...-AcRootParamSet> done at GPS1387878631 - without cmd(0) - nLoad 0, nLoadApply 1, nApplyOnly 0 received from Cm - Load expected at GPS1387878632/1387878631+1 , rtn Ok - setTime 0.000195s
  • The next set of commands could loaded only for GPS1387878633
• To define the  NOISE_CEB_DetTerrace sum using the the noise_colored using a ramp of 1s but without defining the filter for the noise_colored: at this time there is no filter
  • 2023-12-29-09h50m14-UTC>INFO...-AcMatrixChCmSet> NOISE_CEB_DetTerrace - rampTime 1 - 0 noise_white, 1 noise_colored, - loaded and request to apply received - sender "AcousticInjectionCEB"
  • 2023-12-29-09h50m14-UTC>INFO...-AcSumChSet> NOISE_CEB_DetTerrace - rampTime 1s, nStep 8, nb 1/8:  0 noise_line, 0 noise_lineA, 0 noise_white, 1 noise_colored, 0 ENV_PD, 0 triangle, 0 sweep_old, 0 sweep,
  • 2023-12-29-09h50m14-UTC>INFO...-AcEngStateSet> state 3(4) - nload 1/40(0) - nLoadFake 39 - cntLoad 87
  • 2023-12-29-09h50m14-UTC>INFO...-AcCompGpsLoadSet> GPS1387878633/0 - stsGps 1
  • 2023-12-29-09h50m14-UTC>INFO...-AcRootParamSet> done at GPS1387878632 - without cmd(0) - nLoad 0, nLoadApply 1, nApplyOnly 0 received from Cm - Load expected at GPS1387878633/1387878632+1 , rtn Ok - setTime 0.000202s
  • The next set of commands could loaded only for GPS1387878634
• To define filter used to build the noise_colored, to define a new weigth for the noise_colored used by the NOISE_CEB_DetTerrace, ...
  • 2023-12-29-09h50m14-UTC>INFO...-AcFilterChCmFilterSet> noise_colored, filter flt_1004_996, rampTime -1(s) - loaded and request to apply received - sender "AcousticInjectionCEB"
  • 2023-12-29-09h50m14-UTC>INFO...-AcMatrixChCmSet> NOISE_CEB_DetTerrace - rampTime 1 - 0 noise_white, 0.4 noise_colored, - loaded and request to apply received - sender "AcousticInjectionCEB"
  • 2023-12-29-09h50m14-UTC>INFO...-AcRelayChTransCmSet> noise_filter_enbl ask to be set to 1 by AcousticInjectionCEB
  • 2023-12-29-09h50m14-UTC>INFO...-AcRelayChTranSet> noise_filter_enbl, mode set to 1
  • 2023-12-29-09h50m15-UTC>INFO...-AcFilterSet> flt_1004_996(4/0) - applied @20000Hz with frequency warping for channel noise_colored - filter rampTime 1(s) - mode 0  - gain 1.00165 * sos_gain 0.00480853 : fltGain 0.00481644 - resetReq 0 - wk lnSOS 4, lSwap 1
  • 2023-12-29-09h50m15-UTC>INFO...-AcFilterChSet> noise_colored - nameIn "noise_white" - filter prev flt_NONE, new flt_1004_996(0-0) - outFreq 20000Hz, inFreq 20000Hz, wkFreq 20000Hz- slope 1 - slopeRamptime 5(s) - fltRampTime -1(s)- reset "noise_filter_enbl", latch "NONE" - memFlag 0 - outMem 0 - as double_II
  • 2023-12-29-09h50m15-UTC>INFO...-AcSumChSet> NOISE_CEB_DetTerrace - previous ramp not complete remains 0.3126s
  • 2023-12-29-09h50m15-UTC>INFO...-AcSumChSet> NOISE_CEB_DetTerrace - rampTime 1s, nStep 8, nb 1/8:  0 noise_line, 0 noise_lineA, 0 noise_white, 0.4 noise_colored, 0 ENV_PD, 0 triangle, 0 sweep_old, 0 sweep,
  • 2023-12-29-09h50m15-UTC>INFO...-AcEngStateSet> state 3(3) - nload 2/40(40) - nLoadFake 38 - cntLoad 88
  • 2023-12-29-09h50m15-UTC>INFO...-AcCompGpsLoadSet> GPS1387878634/0 - stsGps 1
  • 2023-12-29-09h50m15-UTC>INFO...-AcRootParamSet> done at GPS1387878633 - without cmd(0) - nLoad 0, nLoadApply 2, nApplyOnly 0 received from Cm - Load expected at GPS1387878634/1387878633+1 , rtn Ok - setTime 0.000264s

Comments

• The CM commands are executed in the order in which they are received.
• It seems that the order is not the expected one
• Moreover there is 2 commands to set the weight of the noise_colored used by the NOISE_CEB_DetTerrace.
• the  "load and apply"  is used for each command sent by the  AcousticInjectionCEB server

From my point of view  it would be better to

1. to define the filter   with  "load'
2. to set the noise_white amplitude  with "apply"
3. to enable the filter  : as it is a ACL_RELAY it s done when the command is received
4. to set the weight of the noise_colored for the ENV_NOISE_CEB_DetTerrace with  "apply"

Here attached are the log files of the following injections of low frequency magnetic lines performed with the ENV_MAIN node:

• CEB large coil, lines from 2.5 Hz up to 22 Hz (December 29, approx 3 hours: https://logbook.virgo-gw.eu/virgo/?r=62858 ) (note that the ITF unlocked while injecting the 22 Hz line, the log file has been reconstructed by hand)
• NEB large coil, lines from 2.5 Hz up to 24 Hz (December 30, it started automatically around 7:20 UTC when COMMISIONING mode was set: https://logbook.virgo-gw.eu/virgo/?r=62862, total 3hrs)
• WEB large coil, lines from 2.5 Hz yp to 24 Hz (December 30, it started automatically around 10:2o UTC after the NEB injection finished, total 3 hrs)

Although not so evident in the attached spectrograms, almost all the lines are visible in the Hrec_hoft_2_200Hz channel using 200s or 400s resolution.

Data have to be analyzed to extract coupling functions.

On Dec 31th a set of magnetic sweeps were performed in CEB, NEB and WEB. The sequence was managed by the ENV_MANI node and was repeated twice by setting the ITF in COMMISSIONING mode (see  https://logbook.virgo-gw.eu/virgo/?r=62865 ).  

The first set started at 7:53 UTC and completed at  9:15 UTC. The second set started at 13:35 UTC and completed at 14:58 UTC.

These exact time of each single injection is:

• 1388044411, CEB sweep 1 - 1000 Hz, 9V, 1800s
• 1388046271, NEB sweep 8 - 80 Hz, 1V, 300s
• 1388046578, NEB sweep 40 -100 Hz, 2V, 300s
• 1388046885, NEB sweep 70-1000 Hz, 4V, 600s
• 1388047546, WEB sweep 8 - 80 Hz, 1V, 300s
• 1388047853, WEB sweep 40 - 100 Hz, 2V, 300s
• 1388048160, WEB sweep 70 - 1000 Hz, 4V, 600s
• 1388064954, CEB sweep 1 - 1000 Hz, 9V, 1800s
• 1388066813, NEB sweep 8 - 80 Hz, 1V, 300s
• 1388067120, NEB sweep 40 -100 Hz, 2V, 300s
• 1388067427, NEB sweep 70-1000 Hz, 4V, 600s
• 1388068089, WEB sweep 8 - 80 Hz, 1V, 300s
• 1388068396, WEB sweep 40 - 100 Hz, 2V, 300s
• 1388068703, WEB sweep 70 - 1000 Hz, 4V, 600s

Figure 1 shows that the amplifier of the CEB large coil saturated in the very first part of the injection (from 1 Hz to 8 Hz) reaching the maximum current ~ 6.5 Apk.

Figure 2 shows trend data of the ENV NOISE channels used for our injections. Indeed, all injections were finished when the BNS drop occurred at about 18:20 UTC.

Here attached are the coupling functions measured with frequency lines injected from the large coils. Figures 1 to 3 show the plot for CEB, NEB and WEB buildings. The coupling function is measured from hrec to the modulus of the field measured by the hall magnetometers. Coupling is given in units of m/T. In Figures 3 to 5 to the plot are added tqo reference curves. Data below 10 Hz seem to roughly follow 1/f^5 slope, while data above 10 Hz have a less steep slope which roughly follows 1/f^2.5 shape.

Data points are written in the attached txt files. The formula used for the computation is the usual one described for example by Maria in VIR-0459A-23.