We have made another round of modifications to the ISS second loop board to have a bit of more gain at 1kHz.

Boost 1 zero was increased by a factor of 2, BOOST 2 was added, and a bit of phase compensation for high frequency to allow us to push the UGF to 20kHz-ish (details of the mod will be attached to this entry).

In the attached, black is the measurement from yesterday at 50W, 7dB of VGA gain, boost on.

Blue is today with 13dB of VGA gain, first boost on.

Red is today with 13dB of VGA gain, second boost on.

All in all, the modifications bought us about 13dB at 1kHz.

If we really need more, we should be able to push the UGF even higher, but that needs some serious thinking.

After giving BRSX a day to cool off, I went to EX this morning to see if I could recover it. This required stopping the Beckhoff and recentering the damper assembly, the latter is not trivial. There are a few variables that need to be cleared in the Beckhoff, and they are kind of buried. But, the EX BRS is functional again, commissioners/operators can return the SEI_CONF to the WINDY state.

While I was there, I used a left over iLIGO hockey puck to damp the BRS down enough to re-engage the damping. I couldn't find Krishna's diagnostic plot, but the driftmon overview gives enough information, when the BRS is rung up enough to see on the driftmon (this plot is usually flat, it only goes sinusoidal if the BRS is rung up a lot). For future reference, you can do this by holding a mass above or below the beam, depending on the direction of the driftmon. Watch the driftmon signal and when it starts swinging in a positive direction (when it's at it's largest negative value of the sine wave) hold the mass above the right (-X) part of the horizontal part of the BRS vacuum can. Hold the mass below the can when the driftmon starts moving the other direction.

In order to combat the weirdness with the diode chiller temperature not reaching its desired set point
of 20 degC, I raised the set point to 21 degC.  This seems to have brought the chiller temperature
to ~19 degC.

    Trending the diode chiller temperature, one can see that there's an oscillation in the actual
temperature.  It might be to fool the system we ought to change the temperature set point to 22 degC.

From the trend data one can easily see the temperature oscillations that are apparent when one looks
at the chiller control panel.

Nutsinee, Kiwamu,

This is a belated log.

In this past Tuesday, we went to the HWS table and checked two things in order to study unexplained behavior seen by HWSY (29738). No major conclusion yet.

• We checked the spatial profile of the launching beam using a Thorlabs beam profiler.
  • We checked the beam up to the bottom periscope mirror. No obvious clipping was found.
• We checked the dependency of the returning beam position on the ITMY and CPY angles.
  • This test was unfinished because we ran out the maintenance period.
  • Nonetheless, according to a quick test, the beam(s) the HWSY camera receives seemed to be made of a reflection from ITMY and other reflection from CPY.
  • To be continued.

By the way, we (re-)found that a green beam coming down to the same HWSY path was clipped at its bottom part.

Computed coherences for a 10 minutes period starting at 05:00 UTC, September 22nd.

https://ldas-jobs.ligo.caltech.edu/~gabriele.vajente/bruco_1158555617/

In the low frequency, there's coherence with MICH/SRCL (fig. 1), angular signals (fig. 2) and ITMY L2 control (fig. 3, not sure what is the control topology, maybe this is ok)

One very interesting thing: there is a significant broadband coherence with IMC_DOF_4_P (fig. 4) and with IMC-PWR_IN (fig. 5)

Lines at 57 Hz and 114 Hz are coherent with magnetometers.

We did a TCS tuning test at 50W starting at 6:56 UTC. The CO2X power was decreased from the nominal 400 mW to 0. During the test, we have monitored the frequency and intensity noise couplings to DARM by injecting broadband noise. I will analyze and post the results tomorrow, but it seems like that a small CO2X value is better fore laser noise couplings.

During the test, we tried to compensate for varing optical gain by changing the CAL-CS sensing gain setting. There was a period where we had a 80 Mpc range which was due to uncompensated change in the optical gain induced by the thermal transient. However, we believe that the range will be better with the fine tuned CO2s. Unfortunately, the laser tripped shortely after the test. We did not get a chance to examine the improvement.

We moved the IMC offsets, injecting lines in PIT and YAW on the IMC PZT around 400 Hz, and saw the improvement in the red trace on the second attachment. 

• We watched the lines in DARM and tried to move the IMC offsets to miminize the peaks (similar to what Robert and Kiwamu did in alog 29523). 
• We had tried to minize the jitter to intensity coupling in the IMC yesterday by watching the ISS second loop diodes, but were unable to do better than with no offsets in.  Watching DARM things were much more repeatable and stationary, which makes me think that whatever we are minimizing this way, it is not jitter to intensity coupling in the IMC.  We were suprised that moving the IMC alingment had a repeatable impact on the high frequency noise which is coherent with REFL.  We found that for IMC DOF1 P (which actuates on MC2 using a combination of the WFS signals) a good offset was 170 cnts.  We only tried DOF1 P and Y, so a more precise tunning could be done.  

Tonight we did a few more measurements to try to understand the jitter coupling:

• After powering up, I did some yaw injections on the IMC PZT while Jenne changed modulation depths, which had no impact on the measured coupling, suggesting that the jitter couples to DARM through the carrier field.  We made a reference injection at 01:13:26 Sept 22 2016 UTC, Jenne reduced the 9MHz modulation depth by 6dB (injection at 01:17:35), and then increased the 45 MHz modulation depth by 3dB (injection at 1:24:02).
• We later tried a test of changing the DARM offset, to see what impact it has on our noise, and while doing this did a few more excitations.  Changing the DARM offset had no impact on the measured coupling either, suggesting that jitter (edit: should not be) be coupling to the DC PDs on contrast defect light.
• The coupling for yaw did drop by almost 6dB, (without much change in shape) after the interferometer had been locked for several hours.  This could be consistent with the contrast defect or alingment changing as the interferometer thermalizes. We did not do a before/after thermalizing measurement for pit, but it might be interesting to try this tomorow. While the noise from 200Hz-1 kHz didn't improve dramatically with the thermal change, there are a few peaks that were smaller after the change, (compare blue to purple in the second attached screenshot)
• After moving the IMC P, the shape of the transfer frunction from IMC PZT yaw to DARM changed with the sharp features gone. 

Jenne, Sheila, Kiwamu, Terra, Lisa

We had another long lock at 50W tonight (7 hours so far), and we did several things (list below).

MICH feed forward retuning and IMC WFS offset adjustment had a clear impact on the noise; the IMC WFS offset tuning reduced several peaks in the noise between 500 Hz and 1kHz, and kept the noise above 4 kHz at the minimum. The noise floor is still significantly higher than it should be in the 500Hz-1kHz region, so there is clearly more to do there. Also, unclear at this point how stable these offsets actually are.

Here is the full list of things we did/tried:
Improved MICH subtraction - the best MICH FF gain with the current filters is -15.9 (-13.9 was loaded in the guardian, Jenne updated it) 
IMC WFS offset tuning: Sheila's entry will follow, bottom line is that the noise above 4 kHz is now much more stationary, and we reduced several peaks in the jitter noise between 500 Hz and 1 kHz;
we closed the POP beam diverter (it is still open in normal operation because we are using POPX, we just made a test by opening the ASC POPX loop for a few minutes, from Sep 22, 5.36 UTC to 5.41UTC); we don't think closing the beam diverter made a difference - we opened it again, once we transition to REFL we can close it permanently; 
 we explored different DARM offsets, no real impact on the noise; 
 we run A2L again, it does what is supposed to, but we noticed that the optimal tuning doesn't last long 
 we reduced the ITM ESD bias by a factor of 2 (Sep 22, 5.45  UTC - 5.48 UTC), we didn't see any change in the noise; 
 Jenne explored the dependence of DARM vs OMC alignment by adding offsets to the OMC ASC loops - nothing exciting 
 Terra turned off for a few minutes all the drives for the PIs - no effect on DARM 
 On going: Kiwamu is doing a thermal test by slightly changing the CO2 power - we want to see if we can minimize laser noise couplings this way.
  

Note:  we have a typically big 40.94 Hz ; we found past entries referring to the 40.93 Hz PR2 roll mode; we improved the band stop that it is already in the PR2 M2 actuation path by 40 dB at the exact frequency, but that didn't help; need to follow up on that, the line is big and it is modulated. 

We never had to worry about bounce and roll during the lock, so good job on the tuning  earlier in the afternoon! .

Title:  09/21/2016, Evening Shift 23:00 – 07:00 (16:00 - 00:00) All times in UTC (PT)
State of H1: IFO is locked. Environmental conditions are good. Wind and microseism are low. The seismic band being somewhat elevated in X & Y is subsiding.       
Commissioning: On going commissioning
Outgoing Operator:  Ed
 
Activity Log: All Times in UTC (PT)

23:00 (16:00) Start of shift

Title: 09/21/2006, Evening Shift 23:00 – 07:00 (16:00 – 00:00) All times in UTC (PT)
Support:  Jenne, Sheila, Lisa, Kiwamu       
Incoming Operator: N/A

Shift Detail Summary: Good Shift. Had one lockloss at beginning of the shift. IFO locked for ~7 hours, most at 50W NOMINAL_LOW_NOISE. Commissioning work continues apace.   

Kiwamu, Darkhan,

Summary

EPICS records used for calculation of the DARM time-dependent parameters have been updated with the values calculated using the most recent DARM model configuration parameters for ER10.

All κ's except for κ_PU calculated in the CAL-CS model are within reasonable ranges (see attachment 1):

κ_TST ≈ 0.75
κ_PU ≈ -0.25
κ_c ≈ 1.2
f_c ≈ 345 Hz

We will investigate why κ_PU calculations are incorrect.

Details

The H1 DARM model parameters used for generation of the EPICS values are

• IFO dependent param file: Runs/PreER10/Common/params/IFOindepParams.conf
• IFO dependent, H1, param file: Runs/PreER10/H1/params/H1params.conf
• Measurements param file: Runs/PreER10/H1/params/H1params_2016-09-19.conf

Output files: raw EPICS values, a verbose log and a Matlab file with EP# variables (T1500377-08, Table 2) were committed to the calibration SVN (rev. 3335):

Runs/PreER10/H1/Scripts/CAL_EPICS

./callineParams_20160919.m
./20160921_H1_CAL_EPICS_VALUES.txt
./D20160921_H1_CAL_EPICS_VALUES.m
./20160921_H1_CAL_EPICS_verbose.log
./writeH1_CAL_EPICS.m

Runs/O2/Common/Scripts/CAL_EPICS/writeO2_TDEP_EPICS.m

The changes were accepted in SDF_OVERVIEW (safe.snap and OBSERVE.snap). For ER9 EPICS see LHO alogs 28180, 29104.

Since we have started putting the ALS and SEI_BS nodes into their nominal states, I have uncommented them from the list of monitored nodes for the main IFO guardian. 

Recall (alog 28280) that they were commented out before ER9, and have remained so, so that we could set the Observation Intent bit when the IFO was being left alone, so that the analysis teams could check their pipelines.  Now we're back in the state of being ready to monitor those nodes, I've uncommented them.

As we wind down on creating new nodes, perhaps TJ can do an audit of all the nodes that should be monitored, and make sure that they are being watched by the main IFO guardian.  There shouldn't be anything left that we want un-monitored, except the SDF.  We're still not quite ready for that.

Here is a frequency noise coupling transfer function during tonight's lock.

I calibrated REFL9 into watts using a the factor 2900 V/W (from the diode to the output of the demodulator) and 12 dB, −21 dB, and 2 V/V for the analog and digital signal gains.

If the loop is shot-noise-limited (with 4.6 mW on the diode), this implies a noise of about 40 pW/Hz^1/2. This would imply a noise in DARM that is more than a factor of 10 below DARM shot noise.

[JeffK, Sheila, Jenne]

The bounce mode has been giving us extra trouble today.  In the end, we used a very fine bandwidth spectrum to confirm that it really was ITMY's bounce mode (as the monitor bar graph was telling us).  We tried out some different phases, and ended up with some settings that are now working.  When I went to put the new settings into the guardian, I found that these exact filters had once been used for ITMY, but had been commented out and changed.  I'm not sure how long ago this happened, but it was at least before Sheila made the gain=-1 filters, so that the numerical gain is positive for all test masses, since that filter wasn't included in the commented-out settings. 

So.  Why did the damping phase change some time ago, and why did it change back?  Unknown.

Attached is the OLTF in two configurations with 7dB VGA gain.

50W (actually 48W), AC coupling OFF, and boost ON (solid line)

50W (actually 48W), AC coupling ON, boost OFF (dashed).

UGF is about 9kHz with phase margin of 40 deg or so (boost ON) or 50 (boost OFF).

Boost buys us about 9dB more gain at 1kHz.

I asked Kiwamu to do an analog measurement up to higher frequency on the floor.

I've stepped all four ring heaters consecutively tonight by 4 W each to identify which test mass belongs to the new 17783 Hz/47.7 kHz PI mode (alogs 29820 and 29838). The shift in frequencies of the test masses is dominated by a change in the Young's modulus which we can alter via the ring heater power; by stepping each ring heater and finding which causes a larger relative frequency shift of the mode in question, we can identify which test mass the mode is from.  See T1600080 and LLO alog 18002 for more info. All power changes below were applied to both top and bottom ring heater. All times UTC.

ETMX: 0.75 --> 2.75 @ 9:29 --> 0.75 @ 9:48

ETMY: 0 --> 2 @ 9:49 --> 0 @ 10:08

ITMX: 0.25 --> 2.25 @ 10:08 --> 0.25 @ 10:29

ITMY: 1.25 --> 3.25 @ 10:31 --> 1.25 @ 10:41

Note we just lost lock around 10:42, probably due to these ITM steps. 

Lisa, Kiwamu,

We think that the calibration has been wrong and therefore it lifted up the noise floor below 300 Hz. We need another set of eyes tomorrow to doublecheck our theory (we are currently too sleepy to do systematic investigation).

In short, we believe that the sign of ETMY L3 stage in CALCS (or somewhere else similar to it) is wrong which is exactly the same situation as what happened in this past July (28396).

Because I knew that our DARM open loop model is accurate and consistent with the recent measurement (29748), I made a calibration filter for DARM_IN1 which converts DARM_IN1 to DARM displacement as opposed to the use of both DARM_IN1 and DARM_OUT. Here is a comparison of the CALCS spectrum against the spectrum derived from DARM_IN1.

As shown in the plot, the CAL CS spectrum overestimates the noise floor below 300 Hz or so. This is exactly the same behavior as what we have experienced in this past July. In addition, we have noticed that the noise floor of CALCS changed as a function of the DARM control gain which should not happen in the calibration scheme used in CAL CS. Also, when we flipped the sign of the EY L3 stage in CAL CS, the leve of the noise floor became identical to the one derived by DARM_IN1 and also became insensitive to the control gain. This increased our confidence that the L3 sign was wrong in CALCS.

We are leaving the L3 stage gain flipped (DRIVEALIGN_GAIN -30 --> +30) for the night. If our theory is correct, we regain the binary range back to ~60 Mpc with this change.

Also, we took a DARM open loop and PCAL sweep measurement within the same lock stretch. We did not analyze them yet, but they are available at:

/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/PreER10/H1/Measurements/PCAL/2016-09-21_H1_PCAL2DARMTF_4to1200Hz.xml

/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/PreER10/H1/Measurements/DARMOLGTFs/2016-09-21_H1_DARM_OLGTF_4to1200Hz.xml

Also, my code which generated the calibration filter for DARM_IN1, as well as, the generated filter are available at:

/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/PreER10/H1/Scripts/ControlRoomCalib/CalibrateH1DARM_IN1.m

/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/PreER10/H1/Scripts/ControlRoomCalib/DARM_IN1_calib.txt

Kiwamu I, Jeff K, Darkhan T,

Overview

Today we looked at the calibration line coherence values calculated in the front end. We discovered that in the front-end model with the currently used averaging parameters the coherence value is updated once every 10 seconds (first 1.5 minutes in Fig. 1).

We propose modifying the averaging code to improve coherence calculation and avoid potential aliasing (see details).

Details

A front-end model that calculates coherence uses N data points taken every S_cycles computational cycles to calculate cross-spectral density of the signals. S_cycles is controlled by an EPICS record $(IFO):CAL-CS_TDEP_COH_STRIDE (or simply STRIDE): S_cycles = FE_RATE * STRIDE. There are drawbacks associated with this approach:

• New coherence value will appear every STRIDE seconds;
• Skipping data points will produce aliasing.

We can deal with the first problem listed above by reducing STRIDE. On Fig. 1 we show coherence values for N = 10 and STRIDE set to 10, 5, 1, 1/16 and again 1. Notice that this test was done when the IFO was not locked, thus we do not expect a coherence of ~1. When STRIDE is set to 1/16, we can see that the number of averages must be increased in order to include actual fluctuations in the signals (in the last 2.5 minutes N = 120).

Fig. 1

Several minutes later IFO locked (at t = -5 in Fig. 2) and the coherence became ~0.8. After we set the line amplitude to its nominal value we got coherence of ~1.

Fig. 2

Next, to avoid aliasing we should use all data points (equivalent to setting STRIDE = 1/FE_RATE in the current code) and increase N accordingly. Thus to integrate over 10 seconds and avoid aliasing the current code will require O(FE_RATE * 10) = O(160k) operations per cycle and a buffer for 160k values, which is unnecessary expensive.

A following minor modification of the averaging code can solve the issue described above. Instead of adding every S_cycles'th data point into the averaging buffer, we could insert an average of S_cycles values. E.g. buffer_and_average() can be modified as follows:

...

    static int mini_sum = 0;

...

    //Increment cycle count
    mini_sum += data;
    cycle_counter++;
    if (cycle_counter >= cycles_between_data) {
        buffer[current_pointer] = mini_sum / ((double) cycles_between_data);
        current_pointer++;
        cycle_counter = 0;
        mini_sum = 0;
    }

...

With this modification and N = 160, STRIDE = 1/16 (S_cycles = 1024), the coherence will be calculated with 10 second integration and will require O(160) operations per cycle, and the coherence value will be updated every 1/16 of a second.