We have large glitches when we switch the coil driver state for PRM in every example that I've looked at.  This causes locklosses, about 5 over the weekend.  

One thing that we can do to make this less painfull is to move the switch earlier in the locking process (for example, we can try doing this right after DRMI locks rather than waiting until DRMI on POP).  

A real solution might be to change the front end model so that we can switch each coil separately.   

We could try tuning the delay, as described for the LLO BS 16295

Jordan Palamos, Vinny Roma

We swapped some pem magnetometer power supplies at both end stations to fix the 1Hz glitches mentioned by Robert in his alog 21272.  Since we don't have enough working boxes at the moment, both EY and EX have one of their electronics bay magnetometers disconnected. Specifically EX_MAG_SUSRACK and EY_MAG_SEIRACK (all axes) are disconnected.

Each VEA magnetometer was using a 'new style' power supply / signal conditioning box that was causing the big glitches (these glitches seemed go away when the box unplugged and running on batteries, not sure what the battery life is but I think it would be unfeasable to run as such). At both end stations we replaced these boxes with ones taken from the electronics bay (old style). After swapping, the bad glitches disappeared and the noise floors look much better. The glitchy boxes are now in the EE room.

ECR 1500312, WP 5455

The h1fw0 and h1fw1 computers have been replaced with new computers, which were formerly tested as h1fw3 and h1fw2.  The old h1fw0 and h1fw1 computers have been renamed h1tw0 and h1tw1, and have been reconfigured to write raw minute files to the locally attached SSD RAID.

Both h1fw0 and h1fw1 are now writing science, commissioning, minute trend, and second trend files to SATABoy disk arrays through the ldas gateway computers.

The myricom drivers still need to be updated and configured on h1tw0, h1tw1, h1nds0, h1nds1, and h1broadcast0.

The UPS in the MSR is back in service.  The bypass switch which failed a couple of weeks ago has been replaced.  No power glitches occurred, systems are running normally.

Sep  8 03:23:35 h1conlog1-master conlog: ../conlog.cpp: 301: process_cac_messages: MySQL Exception: Error: Out of range value for column 'value' at row 1: Error code: 1264: SQLState: 22003: Exiting.

Also took the opportunity to reconfigure the replica to allow connections from other hosts. (WP 5481)

Once the series of EQs ceased (2+ hour ringdown), I began the locking procedure.  Alignment looked good so I decided to forego initial alignment and move straight to locking.  A bit of tweaking of the BS in PRMI was all that was required to get the IFO locked, after a few false starts with locklosses at various places on the way up.

10:35 lockloss @ DC_READOUT, ITMx, SR2, ITMy, MC2, and SRM saturated

10:51 lockloss @ INCREASE_POWER, ITMx saturated

11:06 lockloss @ SWITCH_TO_QPDS, PRM saturated

11:13 lockloss @ SWITCH_TO_QPDS, PRM saturated

11:21 lockloss @ DRMI_ON_POP, all TMs and PRM saturated

11:36 locked NOMINAL_LOW_NOISE 70+ MPC

12:27 OMC DCPD and ETMy saturation cause large momentary glitch/drop in range

11:43 set to Observing mode after engaging OMC whitening

13:45 Bubba starts collecting grouting equipment

14:06 ETMy saturation causes large momentary glitch/drop in range

Jeffrey K, Kiwamu I, Darkhan T

Overview

In this alog we present a summary of H1 SUS ETMY UIM coil driver electronics measurements (LHO alog 20846) analysis. UIM coil driver consist of three switchable low-pass filters and a static high pass filter (DCC D070481). According to UIM driver state machine diagram (DCC T1100507) each of the the switchable filters were designed to have a zero at 10.5 Hz and a pole at 1 Hz. A non-switchable filter was designed to have a zero at 50 Hz and pole at 300 Hz.

Fitted zeros of all of the UIM driver low-pass filters were mostly within +/- 0.2 Hz and poles were within 0.02 Hz from designed ( 10.5 : 1.0 ) Hz with uncertainties mostly under 0.3% (see plots below). For individual uncertainties in each of the fit results see tables 1-3.

In this analysis it was not possible to check accuracy of ( 50 : 300 ) Hz pair, because FAST I MONs used for measuring coil driver transfer functions are immune to the effect of this filter, see LHO alog 21142 for more details. This means that UIM driver TF in state 1 should be a flat TF, attachment 4 shows that the TF is mostly flat at low frequencies but we still see something that looks like a high-frequency pole.

Another thing to mention is that we see an unexplained high-frequency feature in UL quadrant measurement that does not appear in LL, UR and LR quadrants.

Details

Similarly to what was explained in our previous alog 21232, the measured transfer functions between excitation and readout signals, apart from driver itself, include also frequency dependent effects from IOP upsampling, digital anti-imaging, analog anti-imaging, and analog anti-aliasing filters shown in the attached diagram (also explained in LHO alog comment 21127).

LP1

LP1 TF was isolabed from all of the other frequency dependencies by taking the ratio of State 2 TF / State 1 TF measurements. Results of LISO fitting of one zero and one pole into the measured TF in a range [0.1 100.0] Hz are given in table 1. Plot of the fitted (model) zero-pole TF vs. measurement and the residuals are shown in the plot under the table.

 Table 1. H1:SUSETMY-UIM driver LP1 fit details 
=================================================
               Fitted LP1 and fit uncertainty  
                        ( z : p ) [Hz]         
-------------------------------------------------
 UL     ( 10.52 +/-  0.06 % : 0.98 +/-  0.06 % )
 LL     ( 10.61 +/-  0.06 % : 0.99 +/-  0.06 % )
 UR     ( 10.34 +/-  0.05 % : 0.96 +/-  0.05 % )
 LR     ( 10.48 +/-  0.06 % : 0.98 +/-  0.07 % )
=================================================

LP2

Similarly to LP1, LP2 TF was isolabed by taking the ratio of State 3 TF / State 2 TF measurements. Results of LISO fitting of one zero and one pole into the measured TF in a range [0.1 100.0] Hz are given in table 2. Plot of the fitted (model) zero-pole TF vs. measurement and the residuals are shown in the plot under the table.

 Table 2. H1:SUSETMY-UIM driver LP2 fit details 
=================================================
               Fitted LP2 and fit uncertainty  
                        ( z : p ) [Hz]         
-------------------------------------------------
 UL     ( 10.50 +/-  0.14 % : 1.02 +/-  0.14 % )
 LL     ( 10.43 +/-  0.12 % : 1.01 +/-  0.12 % )
 UR     ( 10.56 +/-  0.12 % : 1.02 +/-  0.12 % )
 LR     ( 10.55 +/-  0.13 % : 1.02 +/-  0.14 % )
=================================================

LP3

Similarly to LP1 and LP2, LP3 TF was isolabed by taking the ratio of State 4 TF / State 3 TF measurements. Results of LISO fitting of one zero and one pole into the measured TF in a range [0.1 100.0] Hz are given in table 3. Plot of the fitted (model) zero-pole TF vs. measurement and the residuals are shown in the plot under the table.

 Table 3. H1:SUSETMY-UIM driver LP3 fit details 
=================================================
               Fitted LP3 and fit uncertainty  
                        ( z : p ) [Hz]         
-------------------------------------------------
 UL     ( 10.36 +/-  0.33 % : 0.97 +/-  0.34 % )
 LL     ( 10.66 +/-  0.27 % : 0.99 +/-  0.27 % )
 UR     ( 10.60 +/-  0.26 % : 0.99 +/-  0.27 % )
 LR     ( 10.21 +/-  0.36 % : 0.97 +/-  0.37 % )
=================================================

Scripts and Plots

Measurement parameters were committed to calibration SVN:

CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/Electronics/H1SUSETMY_UIMDriver_$(state)_param_$(gps_time).m

The script that loads all of the measurement parameters, calls the fitting function and produces the plots was committed to the calibration SVN:

CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/Electronics/runFit_H1SUSETMY_UIMdriver.m

This script uses the same Matlab functions that were used for PUM coil driver electronics analysis in the same SVN directory.

Plots were committed to the calibration SVN under following names:

CalSVN/aligocalibration/trunk/Runs/ER8/H1/Results/Electronics/2015-09-07_H1SUSETMY_UIMDriver_$(short_description).pdf

CalSVN/aligocalibration/trunk/Runs/ER8/H1/Results/Electronics/2015-09-07_H1SUSETMY_UIMDriver_$(short_description).png

Another EQ in New Zealand got us (and LLO).  We'll see how long it takes to ring down.

LVEA: Laser Hazard
IFO: Locked
Intent Bit: Observing  

All Times in UTC (PT)

23:00 (16:00) Take over from Ed
23:00 (16:00) IFO Locked at NOMINAL_LOW_NOISE, 23W @70Mpc
23:18 (16:18) Commissioners commissioning work while LLO is recovering
23:25 (16:25) Kiwamu – Loading foton files for calibration 
23:37 (16:37) Lockloss - 
23:42 (16:42) Stop at LOCK_DRMI_1F for commissioning work
23:48 (16:48) Lockloss - 
00:00 (17:00) Stop at DRMI_LOCKED for commissioning work
00:05 (17:05) Lockloss – 
01:09 (18:09) Stop at DRMI_LOCKED for commissioning work
02:12 (19:12) Locked at NOMINAL_LOW_NOISE, 22.9W, 74Mpc
02:16 (19:16) Set intent to Observing Mode
07:00 Turnover to Travis



Shift Summary & Observations:

Took over from Ed at 23:00 (16:00). IFO has just relocked at the shift change. Intent bit set to Observing. LLO is down, working on relocking problem. 

Commissioners working on IFO while LLO is down 

Saturations of ETMX at 00:01, 00:02, 00:03 while at DRMI_LOCKED

00:45 (17:45) Reset fiber polarization for Y-Arm to get ALS to lock

02:15 (19:15) 4.5 mag earthquake in Hawthorne, NV – Did not lose lock 
02:50 (19:50) 5.2 mag earthquake near L’Esperance Rock, New Zealand – Did not lose lock
Smooth shift with good locking range. 

 
 

  

J. Kissel, K. Izumi, D. Tuyenbayev, S. Karki, C. Cahillane

We've completed all the to-do list items for comparing all three methods of measuring actuation strength (listed in LHO aLOG 21015). This means that the ER8 / O1 DARM model is virtually complete. Now we just need to compare the full model against measured DARM OLGTFs to confirm no remaining high-frequency systematics in either the actuation or sensing function, the we can declare victory on the frequency domain model side of things. Once victorious there, we
- Update the CAL-CS front-end model to match the low-frequency content of matlab model
- Update the GDS pipeline to matach the high-frequency content of the matlab model
- Generate an inverse actuation filter, and install it in the CAL-CS bank
We hope to complete these items within the next few days.

Results on the Actuation Strength of ETMY
-----------------------------------------
Though there still remain some unexplained systematics, we are confident enough in the PCAL results that we've chosen to use only the PCAL to determine the actuation strength to high precision. The other two methods, ALS DIFF and Free-Swinging Michelson, though less precise, confirm the accuracy within their statistical uncertainty (though rigorous, statistical comparison was not done). The results are as follows:

    'Optic'      'Weighted Mean'    '1-sigma Uncertainty'    '1-sigma Uncertainty'
    'Stage'      '[m/ct]'           '[m/ct]'                 '%'                  
    'ETMY L1'    '5.15e-11'         '2e-12'                  '3.9'                
    'ETMY L2'    '7.3e-13'          '5.6e-16'                '0.076'              
    'ETMY L3'    '1.11e-14'         '1.1e-17'                '0.096' 


Discussion Against ER7, Expectations, & Alternate Displays of Above
-------------------------------------------------------------------
ETMY L1 and L2 are 0.5% and 4.5% larger than what was used during ER7 (see LHO aLOG 18767), as expected because we do not expect this actuation strength to change. The larger percent change on L2 is almost certainly because we've greatly refined our knowledge of the actuation chain electronics. However, both numbers still remain very consistent with models of the transconductance of the coil drivers (the ER8 model uses the cannonical values from G1100968) and actuation stength of the A/BOSEM coil/magnet system (1.74 [N/A] and 0.0333 [N/A] are the fit values for the UIM and PUM used in this model, as opposed to the cannonical 1.694 and 0.0309 [N/A], originally from T1000164).

ETMY L3 is 45% stronger (or 82% depending on whether you choose preER7 or this measurement as the reference) from prior to ER7 (again, see LHo aLOG 18767), we believe simply because the test mass has been discharged. For those who like the numbers reported in more "fundamental" units, the ESD strength has changed from 7.96e-11 to 1.55e-10 [N/V^2].

ETMY L1's uncertainty is so much larger than L2 and L3 because a relatively huge, frequency-dependent systematic still remains in the data. Indeed, if we believe the measurements (and ALL methods show it) the UIM has a rather unsettling right-half-plane zero at around 100 [Hz].

These numbers, in the form of [N/ct], will be added to the CAL-CS model within the next day or so,
    'Optic'      'Weighted Mean'    '1-sigma Uncertainty'    '1-sigma Uncertainty'
    'Stage'      '[N/ct]'           '[N/ct]'                 '%'                  
    'ETMY L1'    '8.17e-08'         '3.2e-09'                '3.9'                
    'ETMY L2'    '6.82e-10'         '5.2e-13'                '0.076'              
    'ETMY L3'    '4.24e-12'         '4.1e-15'                '0.096'

Details & Plots
-----------------
I attached several sets of plots, one set comparing all three calibration methods against the model and each other for each stage of actuation (*_AllMethods.pdf) for all three days of measurement, and the other set comparing all three days of PCAL on one plot for each stage. It is to the latter combined data set that we fit the model and form the uncertainty estimations based on the residuals between that fit and the data. 

The model lives here:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/DARMOLGTFs/
H1DARMOLGTFmodel_ER8.m
with the paramater set
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/DARMOLGTFs/
H1DARMparams_1124827626.m

The comparison script
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/
compare_actcoeffs_ER8.m

uses Kiwamu's recently functionalized
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/ER8/H1/Scripts/
PCAL/analyze_pcal.m
ALSDiff/Matlab/analyze_alsdiff.m
FreeSwingMich/analyze_mich_freeswinging.m

to "quickly" analysis all of last weeks data and to compare against the same model.

One can immediately see that the UIM is the outlier in terms of gross systematics here. I've been trying to chase down the high-frequency discrepanct for days, but in the interest of time, we must move on. Unfortunately, the pre-ER7 UIM measurements were only taken up to the 7 [Hz] upper-limit of FSM method, so we cannot say whether this feature has always been there. Thankfully, because the UIM is well-rolled-off by 100 [Hz] with the hiearchical control filters, even this nasty of a systematic should not impact the DARM calibration above 10 [Hz], given that the UIM / PUM cross-over frequency is roughly 2 [Hz] (we will confirm this more precisely with our model, but it has been confirmed via measurement in LHO aLOG 20941). 

We are using updated electronics chain information for the PUM and TST, based on Darkhan's and my work earlier last week (see LHO aLOGs  21232 and 21189), and this has cleaned up the results greatly from when the data was peviously, individually processed in LHO aLOGs 21049 and 21015.

Evan, Sheila, Jeff B Ed

Earlier today we were knocked out of lock by a 5.2 in New Zealand, the kind of thing that we would like to be able to ride out.  The lockloss plot is attached, we saturated M3 of the SRM before the lockloss and PRM M3 was also close to saturating.  

While LLO was down, we spent a little time on the offloading, basically the changes described in alog 21084.  This offloading scheme worked fine in full lock for PRM, however we ran into trouble using it durring the acquisiton sequence.  Twice we lost lock on the transition to DRMI, and twice we lost lock when the PRM coil state swtiched in DRMI on POP.  Hpwever, we can acquire lock with the new filter in the top stage of PRM and SRM, but the old low gain (-0.02).  We've been able to turn the gain up by a factor of 2 in full lock twice, so I've left the guardian so that it will turn of the gain in M2 (before the intergrator) in the noise tunings step. 

If anyone decides they need to undo this change overnight, they can comment out lines 344-347 and  2508-2514 of the ISC_LOCK guardian. 

Before we started this, the PRM top mass damping was using 10000 cnts rms at frequencies above a few hundred Hz, because of problems in the OSEMS (alog 21060 ).  Evan put some low passes at 200 Hz in RT and SD OSEMINF which reduces this to 2000 cnts rms. Jeff B accepted this change in SDF.  

 The second attached screenshot shows the PRM drives, the references are in the minutes before the earthquake dropped us out of lock.  The red and blue curves show the current drives, with the high frequency reduction in M1 due to Evan's low pass, and the new offloading on.  The last attached screenshot shows SRM drives with the new offloading.  

- Took over from Ed at 23:00 (16:00). IFO has just relocked at the shift change. Intent bit set to Observing. LLO is down, working on relocking problem.
- Commissioners working on IFO while LLO is down.
- Saturations of ETMX at 00:01, 00:02, 00:03 while at DRMI_LOCKED.
- 00:45 (17:45) Reset fiber polarization for Y-Arm to get ALS to lock.
- 02:15 (19:15) 4.5 mag earthquake in Hawthorne, NV – Did not lose lock.
- 03:00 (20:00) IFO locked a NOMINAL_LOW_NOISE, 22.9W, 78Mpc. 

There were six separate locks during this shift, including a new record lock stretch of 25 hours. Typical inspiral range was ~ 75 Mpc. At least two locklosses were caused by earthquakes. Total observation time 57 hours (duty cycle 59%).

Very loud (SNR ~ 1e3) glitches associated with range drops and ETMY saturations continue (roughly 40 during this shift). The correlation between glitches and saturations was investigated; it was found that every SNR > 1000 Omicron trigger (and most with SNR 100 – 1000) was simultaneous with a range drop and an ETMY saturation. Considering the possibility that the 'dust glitches' and the ETMY saturations are actually the same thing. (More detailed alog on this coming soon; details and follow-up at PreO1WorstGlitches wiki).

Spectrograms showed some excess noise in several broadband lines between 10 and 40 Hz on Thursday, and a set of narrower lines between 10 and 50 Hz on Friday. Cause unknown.

The third observation period on Saturday had a significantly higher strain noise floor and lower range (~ 60 Mpc). Robert suggested that this was high frequency noise in an oscillator ( alog). Not followed up.

The periodic 60 Hz glitch which occurs every 72 minutes continues, with slightly lower SNR than during ER7. They are vetoed very efficiently by H1:SUS-ETMY_L2_WIT_L_DQ.

More details can be found at the DQ shift wiki page.

I worked a bit more on the analysis codes that Jeff has written (alog 21015 and alog 21049). I was not able to finish deriving the suspension scale factors. Tomorrow, Jeff and Darkhan will pick up them at the point where I left and will continue working on it.

Major updates:

• DARMOLGTFmodel now incorporates not only each ESD driver circuit measurement but also each PUM coil driver measurement.
  • Darkhan is going to alog some detail of the PUM driver measurement and fitting.
• The ALS Diff measurements are now analyzed using the latest DARMOLGTF model.
• Also, the Pcal measurements are now analyzed using the latest DARMOLGTF model.
  • By eyeballing the resultant plots, I see that the inclusion of the ESD driver and PUM coil driver circuits made the model more accurate. This is encouraging.

Next steps:

• We need to include the latest DARMOLGTF model in the free-swinging Michelson analysis.
• UIM driver circuit needs to be updated. DARMOLGTF still assumes a perfect cancelation between the digital and analog filters.

Some notes:

In the course of the code update, I changed the file organization structure of the ALS diff and Pcal scripts so that they have only one analysis code which can be invoked by specifying a set of data parameters. This organization approach is the same as those for the DARM open loop analysis.

As for ALS diff, the core analysis code is

aligocalibration/trunk/Runs/ER8/H1/Scripts/ALSDiff/Matlab/analyze_alsdiff.m

and the parameter files are in the same directory and named as

H1ALSDIFFparams_20150826.m

H1ALSDIFFparams_20150828.m

H1ALSDIFFparams_20150829.m

As for Pcal, the core analysis code is

aligocalibration/trunk/Runs/ER8/H1/Scripts/PCAL/analyze_pcal.m

and the parameter files are in the same directory and named as

H1PCALparams_20150826.m

H1PCALparams_20150828.m

H1PCALparams_20150829.m

Each parameter file loads H1DARMOLGTFmodel_ER8 with the latest electronics information and subsequently returns the parameter structure, or the familiar variable "par". As usual the parameter structure can be then fed to the analysis code (e.g. analyze_alsdiff.m or analyze_pcal.m) as an input argument. I have not carefully thought about the final return variables from the core analysis codes, but in principle we can write them such that they return the calibrated suspension transfer functions in meters/counts together with some error bars and also perhaps some statistical values. In this way, the final step of performing apple-to-apple comparison between various measurements from various dates can be less painful.

I tried propagating the same file organization structure to the free-swinging Michelson codes, but apparently I am running out my energy for the night and did not finish it yet.

Dan, Evan

This evening we made a qualitative study of the coupling of beam jitter before the IMC into DARM.  This is going to need more attention, but it looks like the quiescent noise level may be as high as 10% of the DARM noise floor around 200Hz.  While we don't yet understand the coupling mechanism, this might explain some of the excess noise between 100-200Hz in the latest noise budget.

We drove IMC-PZT with white noise in pitch, and then yaw.  The amplitude was chosen to raise the broadband noise measured by IMC-WFS_A_I_{PIT,YAW} to approximately 10x the quiescent noise floor.  This isn't a pure out-of-loop sensor, and since we were driving the control point of the DOF3 and DOF5 loops of the IMC alignment channels we will need to work out the loop suppression to get an idea of how much input beam motion was being generated.  Unfortunately we don't have a true out-of-loop sensor of alignment before the IMC.  We may try this test again with the loops off, or the gain reduced, or calibrate the motion using the IMC WFS dc channels with the IMC unlocked.  Recall that Keita has commissioned the DOF5 YAW loop to suppress the intensity noise around 300Hz.

The two attached plots show the coherence between the excitation channel (PIT or YAW) and various interferometer channels.  The coupling from YAW is much worse: at 200Hz, an excitation 10x larger than normal noise (we think) generates coherence around 0.6, so the quiescent level could generate a few percent of the DARM noise.  Looking at these plots has us pretty stumped.  How does input beam jitter couple into DARM?  If it's jitter --> intensity noise, why isn't it coherent with something like REFL_A_LF or POP_A_LF (not shown, but zero)? 

The third plot is a comparison of various channels with the excitation on (red) and off (blue).  Note the DCPD sum in the upper right corner.  Will have to think more about this after getting some slpee.