• Tuesday maintenance will be a busy one, possibly taking all day. Thursday might be used as maintenance as well. 
• ER10 begins end of next week.
• SEI - Everything looks good. Nothing to report
• SUS - Nothing to report. Charge monitoring, ongoing.
• CDS ELE - Intrusive work to be done regarding end station Beckhoff. Ongoing ISS work.
• PSL - Intensity noise investigation to be don this morning
• VAC - Experimentation on CP4 (Tues). Kyle doing RGA bakeout at EX. PT140 can be turned off.
• FAC - Flushing RO system. Punch list items found during safety review to be addressed.
• GC - No internet Tues between 8:00-9:00AM PDT. Final stage of process.

I made a few guardian changes this afternoon. 

• Terra and I found that CHARD blends cannot be on durring the power up, so we are turning them off in DOWN, engage CHARD without them, and turn them on in Low noise ASC.
• I moved the step that switched off the SRM dither and moved it to WFS signals into its own state (SRM_ASC_HIGH_POWER).  Yesterday Jenne and I found that we had to wait about 10 mintes after powering up to do this step, but it has to happen before NOISE_TUNNINGS.  
• Since the ISS seems to have trouble engaging the second loop imediately after the IMC locks, I changed the request in ISC_LOCK DOWN to "LOCKED" and put a request in ISC_LOCK DRMI_LOCKED to set the MC to "ISS_ON"
• I think that I finally fixed the bug that meant that in ALIGN_IFO the dark michelson tried to grab which the optic was still swinging and woud get stuck saturating the BS. 
• In DRMI ASC I added something that makes sure the ASC on switch is on (to avoid self inflicted locklosses)

A few things about ASC

• While moving SRM at high power, I looked for an SRM yaw signal in AS36 I, AS45 I, and REFL45.  It seems like REFL 45 might be the best hope we have, but those signals are all dominated by a loop with a slow oscillation (a few second period), which might be PRC1 or a cross coupling between PRC1 and something else.  If we can fix this, we might be able to find a decent SRM signal in REFL. 
• It seems like the CHARD gain is set too low after low noise ASC, causing it to oscillate.  There might have been a mix up in the guardian, but we should just measure the loop to check anyway.  
• Our DC centerig loops have peaks with high coherence with DARM at a few hundred Hz.  This might be because they are seeing the same jitter that couples to DARM, not because they impose the noise.  

This afternoon Terra and I had a few hours long lock durring which we injected white noise in the ISS, and the IMC PZT (alternating between PIT and YAW). We found that we were able to reduce the jitter coupling by almost 8dB by moving the POP A offset for PIT, which also seems to have reduced the intensity noise coupling slightly (2 dB).  While the couplings did drift as the interferometer thermal state changed at the begining of the lock, the improvements were reproducible with an on off test.  

We started by moving SRM in both pit and yaw, but saw no impact on the jitter coupling that was reproducible with an on off test, which is different from what was seen in alog 25654 and 25657.  The POP PIT offset changed from 0.24 to 0.47.  Durring the move POP 18 improved by about 10%, while the recycling gain improved by about 1%, but neither of these were at their best in the final location. We started trying to move the yaw offset, which looks like it needs to be moved closer to 0, but lost lock due to a bounce mode. 

In the first attachement, you can see a comparison of the DARM spectrum before and after the POP move. The second screenshot shows comparisons of the couplings before and after. The third shows trends of powers durring the change, and the awg set up for the excitation.  

We had two ~12 hour locks Friday and Saturday night and ~3 hour lock today. A new PI rang up in both about 6 hours into lock; 15606 Hz on ITMX, now MODE3. I've added damping settings to medm, StripTool, and SUS_PI guardian. Currently It's set to be damped using the PLL; we'll have to wait for another long lock to see the success. 

Modes 17 & 25 (ETMX 15541.5 Hz & ETMY 15542 Hz) continue to be tricky. Friday night they were both under passive BP (after having PLL jumping problems previously). They rang up about two orders of magnitude before slowly damping (first trend). Saturday night Mode 17 was put under PLL control and there's no large ringup (only a slight upward trend over the 12 hours). Tonight, still in this configuration, I saw the same sudden ring up after about 3 hours. Switching away from PLL to BP only rang up more quickly. Again, with Mode 17 in PLL and Mode 25 in BP, we can keep the overall peak from rising more but I'd like to thwart this sudden rise if possible and know why it didn't occur Saturday night.

Leaving Mode 17 in PLL control with phase settings tweaked (damped it faster this past lock) and Mode 25 in BP. 

PMC has three transmissive output ports: EOM monitor and 2nd loop out of loop sensor are on the same main port downstream of EOM, 1st loop out of loop sensor is on another port, FSS refcav transmission monitor is on yet another port. We looked at these four sensors.

Note that this is not a conclusive entry.

Free running RIN shows all the acoustic features and all sensors agree.

Reduction of power on ISS 1st loop sensors allowed us to look at the free running intensity noise without ISS without saturation, which was impossible before. (Actually 1st loop sensors still saturate once in a while when ISS is off, but not all the time any more.)

In the attached plot of RIN, top panel is without ISS. All sensors agree with each other including acoustic structures (I'm talking about broad bumps at around 440Hz, 580 and 680Hz).

This means that, when free-running, intensity noise coming out of PMC caused by acoustics (jitter-intensity conversion by PMC, maybe some clipping upstream, intensity noise straight out of HPO, everything combined together) is far larger than the acoustic coupling downstream of PMC. But that free-running intensity noise caused by acoustics is reduced to almost nothing with ISS, and as soon as the ISS is turned on we start looking at the residual noise caused by the jitter coupling downstream of the PMC.

We should look at DBB measurements from the past to see if the measured jitter and intensity noise out of HPO are consistent with the above picture.

ISS ON, some differences show up.

On the bottom panel (with the same Y axis scale as the top) is the same set of RIN signals with the first loop on but the second loop off (solid) and with both loops on (broken).

Roughly speaking,

• EOM monitor and the second loop sensor behaves the same way.
• FSS sensor is squashed a factor of 2 more by the 1st loop than the second loop sensor, but doesn't change much by the 2nd loop.
  • (Actually FSS sensor somewhat increases by the 2nd loop.)
• When 2nd loop is turned ON, the 1st loop sensor becomes about the same as the 2nd loop sensor was when only the 1st loop was on.
  • (Actually it becomes somewhat worse than the 2nd loop sensor was.)

Let's define jitter coupling Cn (n=1, 2, E, F for 1st loop, 2nd loop, EOM and FSS sensor) as the sensitivity of a specific sensor to the jitter coming out of PMC. Jitter generated by downstream of PMC (e.g mirror shaking) could also be included in Cn assuming that the new jitter is coherent with the original jitter.

From a simple model that is shown later, we can say that  one of EQ1 and EQ2 below is true, and EQ3 is always true:

• EQ1: C2~CE
• EQ2: |C2|<<|C1|, |CE|<<|C1|
• EQ3: CF~(C1+C2)/2

What we learn.

If EQ1 is true:

There's a common coupling point upstream of EOM monitor and downstream of PMC (e.g. EOM aperture, steering mirrors, PMC enclosure window, back of the PMC mirror).

Local jitter-to-intensity coupling for each of the sensors, e.g. peri mirror causing additional jitter and IMC (with alignment offset) converting it to intensity for the 2nd loop, is much smaller than the common coupling because there's no difference between CE and C2.

We cannot say if C1 is large or small in this case except that |C2-C1| is large.

If EQ2 is true:

Jitter coupling of the first loop sensor path is the large one here (e.g. steering mirrors, PMC output window, back side of the PMC mirror, PD surface).

The EOM and the second loop sensor path jitter coupling is negligible at this level.

EQ3 is always true:

FSS path jitter coupling via REFCAV, AOM etc. happens to be about half of the first and the second loop sensor path added together.

What we should do.

ISS doesn't do anything to the jitter itself. Even if ISS is totally insensitive to the jitter, the jitter still pollutes the IFO. We need to make ISS as insensitive to the jitter as possible, but we also need to mitigate the jitter.

• See if EQ1. is true, and mitigate.
  • EOM alignment, dust particulate on the PMC enclosure window, back of the PMC mirror, and steering mirrors.
• See if EQ2 is true, and mitigate.
  • Dust particulate on the PMC enclosure window, back of the PMC mirror, steering mirrors and BS.
• Reduce jitter. It's clear that there's too much jitter.
  • Flow rate?

If jitter cannot be mitigated, and if the IFO jitter coupling as well as the IFO intensity coupling are both stable, we might be able to sense the jitter (e.g. IM4) and feed forward to ISS. Feeding back to input pointing into IMC by PZT mirror doesn't sound like a good strategy when we need to do it in wide band between 300Hz and 1kHz.

I took a quick look at six hours of 30-minute SFTs automatically
created by FScans this morning (thanks for setting the intent bit!),
to see if recent line mitigation attempts have reduced the amplitude
of the 1-Hz and other combs at low frequencies. Unfortunately, 
the low-frequency noise floor is too elevated to say much yet
about that mitigation success, but in the band 28-32 Hz, where
last night's noise is lower, there do seem to be fewer and lower-level
lines, including reduced amplitudes for the 1-Hz comb with a ~0.25-Hz offset.

Please note that the 56.84-Hz comb reported in the ER9 data
(present but weaker in O1 data)  remains quite strong.
There are other strong lines present, too, but I gather noise-hunting
is just getting started. The 56.84-Hz comb suggests a DAQ problem
somewhere, as noted before.

The attached plots show different bands of O1 data (narrow black curves)
superposed on last night's data (broad red swath). In each case, an 
inverse-noise weighting is used, to suppress periods of elevated noise.

Figure 1 - 10-2000 Hz
Figure 2 - 10-28 Hz
Figure 3 - 28-32 Hz
Figure 4 - 32-50 Hz
Figure 5 - 50-100 Hz
Figure 6 - 100-200 Hz
Figure 7 - 200-1000 Hz
Figure 8 - 1000-2000 Hz

[Sheila, Jenne]

The CHARD error signal blending is back in the guardian.  Note that I engaged it by hand while at NomLowNoise, but it's in the guardian to come on during EngageReflPopWfs.  So, it has not yet been tried during the acquisition sequence. 

Next time, it should just come on correctly with the guardian.  But, in case we need to come back to turning it on later, I used the same procedure for pit and yaw:  Engage the blend filters on the B filter banks, ramp up the gain to 1 over 10 sec.  Once it's started ramping, turn on the matching blend filter in the A filter bank, which ramps on over 10 sec.

In the attached screenshot the solid traces are before blending, and the dashed traces are after blending.  You can see that as expected, the blending significantly reduces the CHARD noise in DARM, and the coherence between CHARD out and DARM decreases dramatically.   There's still a bit of coherence left around 15 Hz with CHARD_Y, but it's much better than it was.

I leave the IFO undisturbed with the Observe bit set starting at 03:10:00UTC.

Sheila Jenne

We made some quick intensity noise injections using the second loop, and jitter using the IMC PZT.  A first look at the data suggests that several of our peaks in DAM which show coherence with the ISS sensor are due to jitter coupling, rather than intensity noise. This menas that we can't fix these peaks with more ISS gain, but we might be able to do something by changing alignment offsets.

The attached plot shows the noise in DARM (taken with MICH FF off, which causes extra noise at 100 Hz), a projection of intensity noise in DARM based on the ISS PDs, and a projection of the jitter noise in DARM based on IM4 trans yaw. I"ve used a coherence cut off of 0.5 for the transfer function measurements used here.  

Still to be done:

• pico IM4 trans so that it is centered in pitch and we can get a better pitch jitter measurement
• Cut out frequencies where IM4 is sensor noise limited (this probably won't affect the peaks of the jitter projection.)
• Repeat with MICH and SRCL FF on to see what we can learn about the noise at 100 Hz. 

I have remeasured the SRCL feedforward transfer function, and put it into the SRCLFF1 filter bank, and into the guardian.

Injections were done: SRCL1_EXC from 22:24:30-22:28:40 UTC, SRCLFF1_EXC from 22:13:20-22:17:30 (no filters engaged, input off, gain=+1).

The matlab file SRCLff.m grabs the data from the injection times, and calculates what the feedforward filter ought to be.  The matlab file Fit_SRCL_Filter.m loads that filter and uses vectfit to find the appropriate poles and zeros. 

The first image (Measured_vs_fitted_SRCLff.png) shows the agreement between the calculated transfer function that we want, and the result that vectfit gives.

The second image (New_vs_old_SRCLff_filters.png) shows the difference between the old FF filter and the new one.

I then took 3 different spectra and coherences, as shown in the third image (NewSRCLff_works.png):  Blue is without any SRCL FF, brown is the old SRCL FF with the gain of -1.3 that Lisa found yesterday, and pink is the new SRCL FF with a gain of -1.  While it's not a lot, there is improvement between 10 Hz and 30 Hz with the new filter over the old.  Either one is better than nothing though.

Note also that the old feedforward scheme was using an elliptic 4.5 Hz highpass (FM8=AC2 in SRCLFF1) while I am using a much more gentle 1 Hz highpass that was already installed (FM3=AC in SRCLFF1) so that I don't change the phase of my filter as much above 10 Hz.  Both cases have the FM6 and FM7 SBVio1 and SBVio2 filters engaged. 

Kiwamu, Sheila, Terra, Corey, Matt, Lisa

We are still dominated by intensity noise, and it is unclear what is the main noise source at 100 Hz right now, but at least we went through mostly all of the other low noise steps, and the locking sequence now works up to the "nominal low noise" state. 

Sheila had put most of the tuning in the guardian already (see  here ); here are some modifications that we had to do to the code to make it run:
 BounceM0 definitions were added to ISC_GEN_STATES
 The settings for the ASC SRC1 loop were loaded, but the PITCH loop was actually not engaged, so I added that (the YAW loop still needs some work, so at the moment it is just left uncontrolled once the SRM dither lines are turned off)
  
 The coil driver switch was only switching one of the coil; we believe that's the reason why we had troubles with violin modes getting excited during several locking attempts earlier tonight. Matt fixed the code (and used "guardian -i" to test it)
 
And this is the list of the parameters that we changed:
 DARM gain is now 700; Kiwamu had changed it from 1400 after finding that the DARM loop had gain peaking, but the new setting was not properly coded in the guardian; 
 DSOFT gains were lowered to deal with a 2 Hz instability appearing in the ASC low noise transition: DSOFT PITCH from 0.5 to 0.2 and DSOFT YAW from 3 to 1 (now in the Guardian); 
MICH and SRCL feed forward tuning (I ended up with -15.9 for MICH and -1.3 for SRCL, gains are still to be fine tuned) 
 

We locked several times tonight and powering up to 50W has been very robust. Parametric instabilities have been taken care of during the whole day (Terra will comment on that, but I believe she has been using standard PI settings without the need of manually adjusting them). However, by looking at the control room monitors from the hotel, it looks like a couple of PI are about to unlock. On a positive side, the summary pages monitoring PIs are working (see for example https://ldas-jobs.ligo-wa.caltech.edu/~detchar/summary/day/20160917/isc/pi_15_5_khz/).

 Before leaving we had left the interferometer in Observing Mode @ 50W, intent bit set to enable detchar monitoring, with intensity noise dominating above 400 Hz. As yesterday, the coupling to DARM increased during the lock. PIs will break this lock. There is more low frequency noise reduction to be done, but it is not immediately clear what is the excess at 100 Hz right now. The calibration shouldn't be too off as Evan checked it recently, at least at high frequency. The range reported by SensMon is above 30 Mpc.. 

(Sheila, Terra, Lisa, Matt, Kiwamu)

SUMMARY:  Took H1 to Nominal Low Noise a few times (albeit briefly), and had issues with violin modes.

Locking Notes:

Tried locking up H1 at the beginning of the shift, but the DRMI & PRMI flashes were fairly non-existant.  Tried a few things to tweak up the alignment on-the-fly, but simply could not get any life with the Michelson.  Spent about 45min trying to get this to work.

Ultimately went for an Initial Alignment.  Had issue with a badly misaligned ALSy.  Tweaked on ETMy AND ITMy for a while before finally getting it.

There was an earthquake midway through the shift, but did not go much over 0.2um/s, so no change made to SEI_CONFIG.

Issues later in the night:

• IMC_LOCK Guardian had issues with the ISS in that it would get in a loop of OPENing/CLOSE-ing the ISS.  The IMC guardian would eventually solve its own problem after a few minutes.  
• Violin Modes for ITMy & ETMy caused us grief whenever we took a 50W H1 past Low Noise (we would make it to Nom Low Noise for a few minutes, but then we'd see the violin modes ring up).

Here is the current status of things in guardian, for people who will be locking latter tonight:

• The violin mode damping is turned off by the coil drivers state, because we found in the past that we couldn't transition the test mass coil drivers with the damping on. As we saw last night, we need to re-engage this for long locks.  This afternoon we tried not turning the violin mode damping on until after doing the coil drivers, but this rung up a mode badly enough to loose lock.  For now I have moved violin mode damping back to 2Watts RF locked. I made the violin state into a state generator, which is used twice, once when we are locked on RF, and again after coil drivers. (The same code gets executed twice).
• In low noise ASC I've added code to turn of the SRM dithers and engage the SRC1P WFS loop, it leaves SRM yaw uncontrolled. This is only partially tested, and starts on line 3488
• NOISE_TUNNINGS can be run as is, but will not engage SRC FF (just type in gain to try it, the nominal value is 1).
• I also added some code to LOW noise ESD ETMY to prevent it from running if the control has already been switched. 

Today I attempted to confirm that the X-end RGA was still functional after having been vented, removed from its protective nipple, re-installed and evacuated yesterday.  I wasn't able to do this due to an inability to connect to it with the RGA-dedicated laptop.  There is some sort of network issue that has been hit or miss for awhile now but, recently, has degraded to the point of being unusable.  I'll try again after investigating and/or at the next window of opportunity.  

Here is the current state of the H1 DAQ.

h1fw0 and h1fw1 have been completely stable for several weeks, and following the code fix on Wednesday 9/7 the frames written by both are 100% identical. These systems have the same hardware and are running the same code, but h1fw0 occasionally asks for retransmissions and h1fw0 never does. The OS install is slightly different between the machines, and we will try cloning fw0 from fw1 next week.

h1fw2 is a front end computer, running U12 with a local disk system for frames. It was running the original daqd code, but went unstable after the Sat 9/10 power outage. We upgraded it to the rcg3.2 daqd this afternoon to see if this improves stability. It is now connected to UPS power.

Now that h1fw2 has the new set of EPICS diagnostics channels I have expanded the DAQ overview medm screen to show these.

Lisa, Sheila, Ed

Durring ER9, the intensity/jitter noise in DARM was worse than inO1, but not as bad as now.  We looked at the PSL accelerometers, and see that the table is moving about a factor of 5 more over a broad range of frequencies from 150-550 Hz. The coherence of DARM with PSL table acclerometers and the ISS PD is also increased compared to July.  The last panel shows the coherence of the table accelerometer with the ISS PD.  Ed trended PSL pressures, and found that the flow rate durring ER9 was 0.5lpm, compared to 0.7lpm now.