Reports 1-1 of 1 Clear search Modify search
AdV-DET (Commissioning)
bersanetti, bonnand, casanueva, gouaty, kasprzack, valera - 22:31 Thursday 21 September 2017 (39537) Print this report
Investigations on OMC angular noise coupling and best SDB1 angular set points

Below is the list of actions performed during the shift of yesterday afternoon (Sep 20). All measurements were taken in low noise 3.

1/ Angular noise injections on SDB1, with bench at "nominal" position:

A first set of low frequency noise injections though the SDB1 actuators was performed while the bench was at its so-called "nominal" position, that is to say the angular position given by the B5 beam drift control, which roughly corresponded to the set point TX = -410 urad and TY = -3559 urad. The list of injections is reported in the table below.

Bench DoF Cut-off Amplitude (in Correction) Start (utc) Duration (s)
SDB1 TX 40 Hz 0.02 V RMS 14:49:33 220
SDB1 TX 40 Hz 0.08 V RMS 14:53:14 480
SDB1 TX 12 Hz 0.2 V RMS 15:02:34 360
SDB1 TY 12 Hz 0.2 V RMS 15:16:18 60
SDB1 TY 12 Hz 0.4 V RMS 15:17:29 60
SDB1 TY 12 Hz 0.8 V RMS 15:18:54 420
SDB1 TZ 12 Hz 0.2 V RMS 15:29:44 80
SDB1 TZ 12 Hz 2 V RMS 15:31:09 300

2/ Scan of the SDB1 set points (TX, TY)

Then we scanned the SDB1 bench set points by +/-5 um around the initial values (obtained with B5 drift control engaged).
 
2.1/ Scan in TX:
 
SDB1_LC_TX  = -410 ==> 15:50:00 UTC  (B5 drift control disabled in TX)
                            = -411 ==> 15:59:30 UTC
                            = -412 ==> 16:00:36 UTC
                            = -413 ==> 16:01:47 UTC
                            = -414 ==> 16:02:26 UTC
                            = -415 ==> 16:03:03 UTC
                            = -405 ==> 16:10:02 UTC
                            = -410 ==> 16:19:22 UTC
                            = -412 ==> 16:25:54 UTC
 
2.2/ Scan in TY:
 
During the scan in TY, the TX setpoint was kept at the value -412 urad which seemed to maximize the optical gain (see explanation below). For this reason the B5 drift control loop had to be kept disabled in TX.
 
SDB1_LC_TY  = -3559 ==> 16:28:10 UTC  (B5 drift control disabled both in TY and TX)
                            = -3554 ==> 16:32:24 UTC
                            = -3564 ==> 16:37:09 UTC
                            = -3558 ==> 16:47:20 UTC
                            = -3557 ==> 16:51:12 UTC
                            = -3558 ==> 16:52:50 UTC
                            = -3557.5 ==> 16:53:47 UTC
                            = -3558 ==> 16:55:29 UTC
                            = -3558.5 ==> 16:58:38 UTC
 
Please be aware that there is a transition phase between setpoints that last between 100s and 10s. Here the reported times are the ones when the command to change the setpoint to the requested value is launched (so this when the ramp starts towards the requested set point).  It might be wise to have a look a the data SDB1_LC_TY/TX when analysing these times.
 
During both scans in TX and TY we checked how the height of the NE and WE calibration lines seen in DARM (in particular the lines at 358.3 Hz and 355.8 Hz) were changing, which can give an indication of the interferometer optical gain. During the scan in TX, we noticed  a small improvement of the ITF response (height of the calibration lines higher by ~2%) for the position TX = -412 urad compared to the initial position (TX = -410 urad). Therefore we decided to keep this position as new working point. During the scan in TY, we also observed some slight variation of the calibration lines but here the optimal position was not so clear, and in any case closer to the initial value.
 
After these observations we added in vpm the possibility to add an offset on the B5 quadrants used to make SDB1 follow the B5 beam :
- SDB1_B5_QD2_H/V_pre       are the raw data from the quadrants.
- SDB1_B5_QD2_H/V_offset   are the offset one can set to change the bench alignment wrt to B5 (in SDB1_Quadrants config file).
- SDB1_B5_QD2_H/V               are the error signals used by the B5 drift control loop, the loop makes it to be 0.

Before the changes SDB1_B5_QD2_H/V were the raw data and error signals of the B5 drift loop as the offset channel did not exist.
Then after relock at LOW_NOISE_3 around 18:20:00 UTC, the vertical offset was set to 25 um (meaning the beam was shifted by -25 um on B5_QD2 vertically, SDB1_B5_QD2_V_pre = -25 um, which corresponded to the bench set point TX = -412 urad). Except for a few minutes between 18:39:25 UTC and 18:43:00 UTC during which the offset was set to 0 um (so the nominal condition for the alignment of SDB1).
 
3/ Angular noise injections on SDB1, with 25 um offset on B5_QD2_V
 
We performed two types of broad band noise injections with the SDB1 bench set at the new set point (with the 25 um offset in B5_QD2_V): some injections at high frequency (flat noise above 100 Hz) and some injections at low frequency (below 12 Hz).
 
3.1/ Noise injections above 100 Hz
 
Bench DoF Amplitude (in Correction) Start (utc) Duration (s)
SDB1 TX 5.5e-5 V RMS 18h58m31 120
SDB1 TX 2.7e-4 V RMS 19h00m40 100
SDB1 TX 2.7e-3 V RMS 19h02m23 180
SDB1 TX 1e-2 V RMS 19h05m48 300
SDB1 TX 5e-2 V RMS 19h11m03 140
SDB1 TY 5.5e-3 V RMS 19h23m52 60
SDB1 TY 2.2e-2 V RMS 19h24m55 120
SDB1 TY 5.5e-2 V RMS 19h27m14 210
 
3.2/ Noise injections below 12 Hz
 
Bench DoF Amplitude (in Correction) Start (utc) Duration (s)
SDB1 TX 0.2 V RMS 19:37:35 160
SDB1 TX 0.1 V RMS 19:40:18

180

 
 
4/ Angular noise injections on SDB1 with large misalignment
 
At 20:00:00 utc we changed the offset in B5_QD2_V from -25 to -180 um which moved the bench from the position TX = -413 to TX = -419. This made the ITF lock quite unstable and Diego had to adjust offsets in the alignment loops to compensate this SDB1 misalignment. We then started to inject a low frequency noise (cut-off at 12 Hz) in TX at 20:26:49 utc, amplitude 0.1 V RMS. The noise injection lasted a few minutes until the ITF unlocked.
 
We stopped the activity here. We decided to put back the offset of 25 um in the B5_QD2_V signal of SDB1 and leave it like this. It is not clear if this offset (which was found to correspond to the optimal position of SDB1 in TX for what concerns optical gain maximization) is going to be stable in time. Therefore, one useful test to do in one of the next shifts would be to perform a scan of the OMC1 temperature with the beam directly reflected from north arm in order to check that the OMC alignment is still correct with this offset.
 
Analysis of these data to be followed.
Comments to this report:
bersanetti, ruggi, gouaty, bonnand, frolov, kasprzack - 19:43 Friday 22 September 2017 (39554) Print this report
We wanted to investigate if the 270 Hz and nearby noise peaks are from the OMC beam jitter.
The first plot compares the best reference from last week (when the 270 peaks were very low) to the current typical noise floor and to the two bench misalignments (TX, TY). In one of the misaligned state (TY), the 270 peak height increased as if it was driven by the beam jitter. But we didn't observe the same increase for the other misalignments - TX or TY.
The second plot shows the response to excitations from 100 to 400 Hz. A different set of peaks was excited while the height of 270 peaks remains unaffected. This would mean that we are not only looking at OMC beam jitter coupling, but also at some other noise coupling that gets affected by the global alignment (by moving SDB1, we are also moving the beam on SDB2, which has QPDs used for common arm alignment).

Furthermore, the DARM coherence with PRCL at 270 Hz is high according to BRUCO (see third plot). This makes us suspect that the peak is actually from frequency noise coupling, and it gets modulated by the global alignment.

The 270 peak was persistently in the DARM spectrum starting from September 15. Prior to this date, there were several days when the best spectrum of the day didn't have the peak. Coincidentally, the injection lab AC status changed on September 14.

During the same shift, we tried to see if SPRB alignment had an effect on the peaks, but scanning the full SPRB_B4_112 and LSC_B4_DC ranges didn't modulated the 270 peak.
Images attached to this comment
genin, fiori - 11:01 Saturday 23 September 2017 (39564) Print this report

Regarding the 270 Hz peak which seems to be present on DARM since the Laser room to SIB2 room door and cleanroom to SIB2 door have been opened. We can clearly see it on the 

the Beam Pointing control system signals (see attached plot). We could not see any excess of accoustic noise at that frequency but there is a slight effect on the seismic noise measured with the accelerometer on top of the EIB which suggest that a resonance of a mount (maybe EIB_M6 one) could be excited (see entry https://logbook.virgo-gw.eu/virgo/?r=38640).

The strange thing is that the excess of noise read by the BPC quadrants is very small and we are not expecting any limitation from input beam jitter noise at that level when looking at the last noise projection we made on the 13th of september (https://logbook.virgo-gw.eu/virgo/?r=39459). At that time, we injected a large boradband noise without being able to see any coherence for frequencies higher than 100Hz.

Maybe this noise comes from the CB clean rooms and couple the ITF through PRCL (SIB2 PDs)?

We plan to make more investigations next week on that issue trying to understand it better and solve it.

Images attached to this comment
Search Help
×

Warning

×