Seeder laser follow-up (PSL-LAS-0101-03)
Conclusion:
-> it seems the seeder is quite stable and can't beresponsible for any power drop at the neoVAN output
data taken from logbook since December the 17th:
171212 (17 December 2017)
# Calibration LB_M6 (Seeder)
17,1 W reflected from LB_M6 to neoVAN = 17,3 mW transmitted by LB_M6
--> T_LB_M6 = 1.01e-3 TBCompared with 1.084e-3 given by the LMA
(let's assume that the LMA calib is correct since, I don't remember having zeroed the calo)
171218
T_LB_M6 = 18,2 mW
--> P_neoVAN_In = 16,8 W
## Slave current change to check whether we can get 20 W at neoVAN_in
D1/D2 (A) P_LB_M6 (mW) -> P_neoVAN_In (W) = P_Slave after LB_L5
29.4/30.3 18.2 16.8
30.5/30.3 18.2 16.8
30.5/31.4 18.2 16.8
31.5/31.4 17.8 16.42
30.5/32.4 17.6 16.23
180612
# measure of the seed power : 16.6 W (~ 18h LT)
Tests around the fibers bringing the pump to the neoVAN
# we check whether that the swap of spare fiber FOPT-1801-01 with FOPT-1703-02 is mandatory
- with the fiber PSL-FOPT-1703-02 on the neoVAN, we get PMC_TRA = 66.0-66.7 W
- we install the fiber PSL-FOPT-1801-01 instead of PSL-FOPT-1703-02, we get PMC_TRA = 65,5 W
we come back with fiber PSL-FOPT-1703-02 on the neoVAN, we get PMC_TRA = 66.0 W
--> it seems both fibers are equivalent as long as some bending is put on PSL-FOPT-1703-02
# Fiber bending (for stress) location
we check whether the bending set near th eamplifier head is still mandatory while another bending has been set on the diode side
with stress near the head released: the power drops and comes back to 66 W
-> seems the stress becomes useless as long as another stress has been set near the diode box
Rem. : removing the stress on the fibers at the diode side give a drop of 1 W at the PMC output
-> it seems cumulative bending of the fiber does not bring any significative improvement. One single beding (one fiber turn over a 10 cm diameter round) seems enough
!!! I ve checked that PD_PMC_REF is well centered !!!
AOM disconnected
d (m) P_M9(W) TEM02_V TEM02_% TEM00_OR(V) C_R TEM00_R PMC_TRA(W) Comment
0,64 89.0 0.020 0.5% 3.45 0.156 0.56 66.2 Step180613_1
0,62 90.3 < 0.02 0% 3.46 0.150 0.52 67.8 Step180613_2
0.60 89.2 <0.02 0% 3.42 0.155 0.53 65.9 Step180613_3
-> seems d=0,62 m is optimum
0.60 93.2 0.050 1.4% 3.54 0.169 0.65 68.3 Step180613_4
0.62 90.7 x x x x x x Step180613_5
0.62 95.1 0.180 5 3.58 0.179 0.820 63.5 Step180613_6
0.62 90.6 < 0.02 0 3.40 0.158 0.540 65.7 Step180613_7
0.62 89.1 x x x x x 66.3 Step180613_8
0.60 89.4 < 0.02 0 3.425 0.147 0.504 67.2 Step180613_10
-> seems d=0,60 m is better than 0,62 !... and we can't recover what we got in Step180613_2 (see Step180613_7)
--> we stick to d = 0,60 m
# Play with different pump
0.60 x x x x x x 69.1 Step180613_11
0.60 x x x x x x 68.5 Step180613_12
0.60 x x x x x x 68.9 Step180613_13
0.60 x x x x x x 69.1 Step180613_14
0.60 x x x x x x 67.0 Step180613_15
0.60 x x x x x x 69.4 Step180613_16
0.60 x x x x x x 69.9 Step180613_16_bis
0.60 93.0 0.06 1.68% 3.56 0.620 0.157 70.2 Step180613_17
0.60 x x x x x x 68.4 Step180613_18
0.60 x x x x x x 69.9 Step180613_19
-> seems Step180613_17 is the best config
!! the meas. of the TEM02 mode is missing... should be done any time later
This may sligthly change the conclusion on the most efficient pumping configuration
indeed:
- we expect Step180613_12 to lead to 5% would lead to 72 W assuming TEM02 is cured (same config as Step180613_4)
- Step180613_17 should present even more TEM02 (TBC) since the pump config is unbalanced is worsen (4,5/5,5 A) could lead to 73,7 W assuming TEM02 is cured. Again: To Be Checked
Legend:
d: dist(LB_L5-neoVAN input side)
Step180613_1: 4.2/5.3
Step180613_2: 4.2/5.3
Step180613_3: 4.2/5.3
Step180613_4: 4.7/5.3 Pgoes to 94.2 W after 30'
Step180613_5: 4.2/5.3 for a double check
Step180613_6: 4.7/5.3
Step180613_7: 4.2/5.3
Step180613_8: 4.2/5.3 before moving to 0,60 for double check that this is the optimum
Step180613_9: 4.2/5.3
Step180613_10: 4.2/5.3
Step180613_11: 4.5/5.3
Step180613_12: 4.7/5.3
Step180614_13: 4.6/5.3
Step180615_14: 4.4/5.3
Step180615_15: 4.5/5.1
Step180615_16: 4.5/5.4
Step180615_16_bis: 10' later after realign
Step180615_17: 4.5/5.5 (by the way, we got also 67.5 with AOM), 18:00 LT
Step180615_18: 4.6/5.6
Step180615_19: 4.4/5.4
--> we choose the config Step180615_17 from now
it leads to 70,2 W at PMC_TRA with the AOM disconnected
it leads to 67,5 W at PMC_TRA with the AOM back connected
We have swapped the mirror mount of LB_M6 and LB_M7 from newport ones (picture 1) to Polaris_Thorlabs ones (picture 2). The newport are suspected to be unstable since they show some "hysteresys" when moving the micrometer adjusters forward and backward.
We 've checked it did not impact on the overall power.
This should help in long term stability although it is unlikely that this will cure the neoVAN beam drift mentionned in 41742
Here is the text of the original logentry with a readable format:
" it seems the beam at neoVAN output continue to drift in horizontal direction, see figure1_plots 5&7
little drift in the vertical direction (figure1_plot 6&8)
This is confirmed by the camera installed at neoVAN output (log entry 41598),
The camera installed on the seeder (before the neoVAN input) states also an horizontal drift (figure1_plots 1&3)
This trend is a constant drift with no sign of asymptotic convergence since ~ the seeder waist size adjustment (4/04/18), see figure2: the line dark-brown (refers to PSL_AMPL_QD_FF_H) "
# Monday 11/06/2018: seeder size status before neoVAN insestigation with neoLASE engineer
- we set the neoVAN seeder size to 131 um, located @ ~ 6 cm upstream the neoVAN input side
it should be compliant with the nominal setting from the datasheet: 130 um at 5 cm (see attached pdf doc) .
- Slave-to-neoVAN telescope tuning:
we set the distance between LB-L4 and LB_L5 to 33.1 mm (lens center -to-lens center)
dis(LB_M5'-LB_L4) = 380 mm, dist(WP5-M5') = 535 + 270 mm
the beam profile measurement of the seeder is stored into 180611_SlaveTelescope_BeamProfile_r.txt, (see also plot). The profile of the seeder is got in transmission of LB_M6 mirror
--> that is 131 um at 55 cm from LB_L5 (from propagation measurement) and 6 cm before the neoVAN input side ( we set dist(LB_L5,neoVAN) to 61 cm)
The ITF lose lock yesterday night (see pic 1), and after some attempts got stuck by ITF_STATUS autorelock. I tried to restore the lock, but apparently the 7Hz is too high even for the simple LOCKED_ARMS status (see pic 2).
I leave the ITF in DOWN, to be checked later.
This morning at 05:01 UTC, I recieved a DMS notification for data loss.
I found the VPM processes FbmMain and Rds_20 crashed, so I restarted them at 05:07 UTC.
13:00 ITF locked in LOW_NOISE_3 with Horizon 31 Mpc.
The planned Activity on ISC went on without major problem from 13:00 to 16:50 UTC.
Commissioning Mode Stopped.
Calibration Mode Set.
From 16:57 to 17:27 I performed the planned Calibration "CALIBRATED_DF_SENSITIVITY".
Calibration Mode Stopped.
17:50 ITF unlocked
17:50 UTC I went at WE to switch off SWEB Step Motors.
18:00 UTC I went to DetLab to switch OFF SQZ particle counter inside acoustic enclosure. The DET particle counter also was very noisy. I switched OFF also that one. (Monday morning it can be swith ON again).
The effect is clearly visible in the Spectrograms (fig1) and (fig2) but not in the DMS plot (fig3).
ITF locking has been soon restored. At 18:58 UTC the Horizon reached 33 Mpc.
21:00 UTC ITF left locked in Autorelock Mode.
Suspensions
Under request of Ruggi, at 20:50 UTC, I changed the Gains of ADD UNLOCK from 0.1 to 0.2 for NE, WE, NI and WI F7 CoilDrivers via damp card.
DMS
14:45 UTC under request of Soldani I changed the thresholds for INF_NEB_CHILLER_TE_IN, INF_NEB_CHILLER_TE_OUT.
18:10 UTC ParticleCounterSQZ subflag muted till monday morning.
The ITF lose lock yesterday night (see pic 1), and after some attempts got stuck by ITF_STATUS autorelock. I tried to restore the lock, but apparently the 7Hz is too high even for the simple LOCKED_ARMS status (see pic 2).
I leave the ITF in DOWN, to be checked later.
At 17:50 UTC I switched OFF the SWEB Stepmotors Power Supply crate.
At 18:00 UTC I switched OFF the SQZ Particle Counter inside the SQZ enclosure.
BS_TY has been closed again full bandwidth on B5_QD (fig 1), thank to the good control filter developed by Ettore. It has been decided to reduce the offset, going back to a higher CMRF (fig 2), but obtaining a better behavior of the sidebands (fig 3) and the AA signals in general. When the old offset has been tried with BS closed, the ITF unlocked.
Many other small adjustement have been performed, in particular the dither lines have been further reduced and all the vertical dampers have been engaged. The effect on DARM at low frequency is visible in fig 4.
A standard set of noise injections on the longitudinal DOFs (SSFS excluded) have been performed; the data is stored in /virgoData/NoiseInjections/LSC/LSC_injection-1213108483.txt.
A large peak at 54 Hz has appeared in DARM; it was not large the morning of Wed June 6th or anytime before, shows up fully on Thursday and has remained since. (We locked very briefly Wed and the peak can be seen but is not very high). It seems due to some change occuring on Wednesday. There was a new chiller turned on Tuesday for a test, but was turned back off as evidenced by the accelerometer close by. There was cabling change of PD for IMC. I have looked through most accelerometer, seis, and mic channels and do not see evidence of anything mechanical turned on during this time.
Regarding the squeezer bench extra noise I think this periodic noise is coming from the particle counter installed on top of the squeezing bench plexiglass enclosure.
It would be wise to switch it off.
I was notified by Francesco Berni about DMS warning about "high seismic noise on EDB" (Figure 1). As shown in the plot the EDB seismic noise had a slow rms increase starting from June 3 then, on top of it since June 9 afternoon there are transient seismic shakes with 2-4kHz frequency content. The first of thee shakes seems occurred on June 9 at 13:52 UTC. They seem local to the EDB since nothing is seen on SQZ bench, also there is little acoustic content (Figure 2). Beside this there are periodic (10min approx) acoustic (mainly) transients heard by the mic on the EDB bench (Figure 2 as well). I could find no correlation with HVAC signals for neither of them.
I wander if something has been added or switched on the bench (some piezo?) and if shakes can be associated with ITF operation?
While looked at it I noticed that both the accelerometer and microphone on the nearby SQZ bench see loud noise transients with 2min rate (in Figure 2). It must be some device on the bench.
Regarding the squeezer bench extra noise I think this periodic noise is coming from the particle counter installed on top of the squeezing bench plexiglass enclosure.
It would be wise to switch it off.
At 18:00 UTC I switched OFF the SQZ Particle Counter inside the SQZ enclosure.
During the lock of the night we could see an high residual power noise of PSTAB (see attached plot), experts have been informed.
The planned activity on ISC went on for all the shift without problems; it will continue also in the first part of the afternoon.
Looking at spectrograms the extra noise bump on B8_DC started on Monday morning (figure 2).
This coincides with the turning on of the SWEB motor crate to allow the recovery from the west end building air conditioning problem.
It would be good to switch-off this motor crate if it is no longer needed.
At 17:50 UTC I switched OFF the SWEB Stepmotors Power Supply crate.
Thi smorning we removed the individual demodulation phases per sector of the quadrants on B1p to make some tests.
First of all we have injected a line @ 2Hz in WE and BS
| GPS | |
| WE tx | 1213080555 |
| WE ty | 1213081660 |
| BS tx | 1213081839 |
| BS ty | 1213081375 |
We also re-tuned the phases of both quadrants but we basicaly see the same signal in both by eye, see FIgure 1. We will post a more detailed analysis with numbers but it is quite clear that they are very similar.
Due to my lack of familiarity with DSP cards, I did not realize that I did not put a 0 on the right BS AA TX switch. Paolo showed me that I put a zero on the switch related to BS TX drift control (page 2 of BS Mar card), while the switch for the full bandwidth was on page 3 of the same card. The switchs for the BS full bandwidth were already both set to zero, so even by enabling the general switch in metatron code does not actually engage the loops.
Bottom line, the sentence:
at 22:31 LT we engaged the full bandwidth loop for BS, but only on TX, manually disaabling the TY one on the DSP card
should read:
at 22:31 LT we disabled the drift control loop for BS, but only on TX, manually disaabling the TX on the DSP card
Fomr GPS=1211912094 to GPS=1211915094, with the ITF in free short michelson configuration, we have performed a ramp of the phase offsets SDB2_B1p_6MHz_phi0 and SDB2_B1p_8MHz_phi0 in order to study the best way to determine automatically, before each FREEMICH calibration, the value of phi0 which maximizes the signal SDB2_B1p_PD1_{6,8}MHz_I and minimizes the signal SDB2_B1p_PD1_(6,8)MHz_Q.
The ramp was from -pi to +pi with a step of 2*pi/100 every 30 sec.
Plots 1 and 2 show, for the 6MHz and 8 MHz signals, the value PHI = Mean(arctg(Q-mean(Q)/I-mean(I))) computed over 30 sec (left plot) and the range of the signals I and Q (right plot) during this phi0 ramp.
First conclusion is that the value PHI gives the phase offset to be applied in order to get the largest range on I.
But, for the 6 MHz signals, if PHI is around zero, there is an ambiguity and we need to know the sign of the slope of PHI versus phi0 in the ramp to separate the cases "I range is maximum" and "Q range is maximum".
Plot 3 shows the I and Q signals resampled at 0.03 Hz and shows that there is a permanent offset on the signal SDB2_B1p_PD1_6MHz signals which is not present in the 8 MHz signals. The 6MHz signals seem contaminated by some noise that makes its mean value depend on the phi0 used.
Plot 4 shows that the PHI formula works also for the 56MHz signals (where no phi0 ramp was done). We can check on this plot that the phase offset currently used is the correct one.:B1p_56MHz_phi0 is around 1.2 rad and PHI is around 0 and the range of the I signal is maximum with respect to range of Q signal.
PHI formula will be implemented in CALI METATRON soon. But it would be worth to understand what makes the difference between the B1p_PD1_6MHz and the B1p_PD1_8MHz signals
A zero in the DSP channel was forgotten after the investigation concerning the driving noise.
I have repeated the same analysis performed in 41814 using the channel of this entry and only few channels show a stable time constant, thus only few channels show a real transient after each lock see Figure 1 (I have considered only channels with time constant < 3e4s).
As it is pointed out before the 56 MHz phi has a time constant longer than the DC power but much shorted with respect to the cavity mirrors. Moreover its time constant is close to the 6 and 8MHz EOM signal decay.
A zero in the DSP channel was forgotten after the investigation concerning the driving noise.
>> we set up a quick workaround for the LNFS bug reported recently; we will see in the next unlocks how it is working; the idea is to have a timer after which the reset command is sent again;
This problem should be solved by integrating the LNFS commanding inside the automation node itself instead of living on a separate script. This should eliminate concurrent behaviour. Tests could be made during coming automation dedicated shift next week.