A while a go RichM gave me some new blend filters to try. Because we have no IFO, I've installed them on ITMY in the X and RY dof's. I'll leave the ISI like this overnight to get a good long quiet stretch to compare data with.So far, not much to say. I'll put more detail in tomorrow.

I used the sitemap to look at the code revision numbers of each timing device. All DuoTones, Master/Fanouts, and Comparators are programmed to the correct code revision (documented at https://awiki.ligo-wa.caltech.edu/aLIGO/TimingFpgaCode). The 3 issues are:

1) The C_MA_A IRIG-B in the MSR shows Rev. 0. This is a known issue and will be corrected by reprogramming the IRIG-B when possible. The IRIG-B seems to be producing the correct time code output, so we're going to test the effect of a temporary (< 15 minute) interruption to the IRIG-B signal on an IRIG-B device in the test stand to see whether we can reprogram the MSR IRIG-B without causing hiccups.

2) The other 2 IRIG-B in Y_FO_A and X_FO_A show Rev. 133. They should show Rev. 113. They are otherwise operating nominally. It is possible that the timing wiki needs to be updated, or that the revision number was entered incorrectly in the firmware source code.

3) All of the Oscillator Locking slaves in Y_FO_A, X_FO_A, and C_FO_A show Rev. 83, which is the deprecated code. They are operating nominally. Again, this could be an incorrect revision number.

We need to check whether 2) and 3) are meaningful problems that need correction.

Note, as a further contamination test, we blew the N2 gun across a new 4" witness wafer for the same amount of time/distance as we did during the discharge and inspected.  We blew for 30sec at 20psi 2" from the wafer.  Upon inspection, we found the wafer still looked very clean, as expected.

We have been stating the witness wafers as 3" wafers, but in-fact they are 4" wafers.  There are not 2 sizes on site.

Today we finished up the ETMx in-chamber activities adn closed the door.  Sequence of events:

Filiberto checked for TMS QPD cable grounds at the feedthru.  All ok there. (Actually did this yesterday afternoon.)

Kyle and Gerardo finished the port work (that was scheduled in the original vent plan).

Fil check the ESD pinout with aid from an "inside" guy multiple times in order to make clear the understanding of the pins on the air-side of the feedthru.

Jim unlocked the ISI.

Performed the discharge procedure on ETMx as per T1500101 - again, this took around 1 hour.

Swapped the vertical witness plates (1" and 4") that were mounted on the QUAD structure.  1"  optic vertical on floor is now: SN 242.

Removed all equipment and re-suspended lower masses.

Dust counts taken throughout.  Counts were nomnally 30-200 counts in all sizes of particles with few bursts up to 500 in th 0.3um size once in a while when 2 or more people were moving in chambelr.  Counts were zero or 30 (max when entering chamber after lunch or in morning.

Measured ETMx main and reaction chain P and V TFs.  All looked healthy with purge air on and soft cover over door.

Door being put on.

Checked EQ (earthquake) stops (Note this is the same as we found/set on ETMy):

Barrels of ERM ~2mm

Barrels of ETM ~1.5-2mm

12 o'clock stop ~3+mm

PenRe barrel ~1.5-2mm

PUM barrel ~2mm

All face stops 1-1.5mm

ETMX ESD

1. Continuity test was done on each pin to verify each section LR, UR, BIAS, UL, and LL. 
2. The old vacuum feedthrough was removed by Gerardo
2. The in-vacuum cable was re-terminated for installation of the new UHV 5-way coaxial connector.
4. In-vacuum cable was connected to feedthrough
5. Gerardo installed new feedthrough.
6. In-air cable connected to feedthru.
7. All pins were HIPOT tested to 800V and passed.
8. Pins were rechecked, by grounding each section to ground (in chamber) and verifying at feedthrough (air side).

 Pin layout for ESD (@ Feedthrough) is as follows:

   Pin 1 - LR
   Pin 2 - UR
   Pin 3 - BIAS
   Pin 4 - UL
   Pin 5 - LL

Following E1400430, Pin 1 is slightly offset in spacing from the other pins.

Removed 4" wafer from HAM6, (T1500311).

Took PET swipe from south side of chamber from base of OMC to edge of the ISI table.

Both samples have been prepared for processing at CIT.

Note: The placement of the 4" wafer was in a difficult spot to remove. One had to lean into the chamber on an off balance stance and remove the wafer from the center of a group of small optics. There were no good handholds or bracing points available. The wafer being placed flat on the ISI table had to be pushed across the table until it came up against a cable clamp before one could get the tweezers under it. There was a risk of bumping or disturbing the optics around the disk. See attached photo. 

Dust counts taken during work, inside and outside the chamber were all zero or in the low 10s of counts.

We found an error in the GDS calibration filters originally generated for ER7 and discussed in LHO alog #18813.  Originally, we had double compensated for the AI filter in the actuation chain and the AA filter and OMC whitening poles in the sensing chain.  This was corrected on June 4, 2015 (at GPS time 1117505816) and new plots of the filters are attached.  The new filters file is located in the calibration SVN: aligocalibration/trunk/Runs/ER7/GDSFilters/H1GDS_1117505690.npz.

Particle counts taken during the removal of the HAM6 south & east doors on 06/17/15. Counts were taken with an older HPPC-6, using differential mode.

   HAM6 South Door 
Time	0.3µ	0.5µ	1.0µ	Sample Location
13:24	100	0	100	In cleanroom – breaking bolts
13:32	300	100	100	In cleanroom – breaking bolts
13:35	0	100	0	In cleanroom – breaking bolts
13:43	40	20	20	Top of door flange venting purge air
13:48	3050	250	40	LVEA Room air

   HAM6 East Door
Time	0.3µ	0.5µ	1.0µ	Sample Location
14:07	2530	750	320	LVEA room air
14:09	960	350	320	In cleanroom – breaking bolts
14:30	0	0	0	In cleanroom – Door coming off
14:43	20	10	0	In cleanroom - Putting on the door covers
15:00 	10	0	0	In chamber – Cover on
15:39	50	20	10	In chamber – Cover on

Discharge of ETMy WBSC10 (in chamber)

Using the ETM / ERM Gap discharge procedure https://dcc.ligo.org/LIGO-T1500101 we discharged the optics in chamber at ETMy. Remember that this procedure includes a swing back of the ERM i.e. an increase in the gap between the optics. Using the electrometer https://www.trifield.com/content/ultra-stable-surface-dc-volt-meter/ mounted centrally (on an X brace) 1" from the back of the ERM (i.e. between the ERM and TMS) we measured the following voltage: -

Note:- The humidity at 10:00 am was 30%.

Note: - Prior to starting we zeroed the electrometer (11:12 am) with the cap on.

1) At start (after electrometer was mounted 1" from back of ERM and prior to swing back) = 6.8 V (this was steady over couple of minutes).

2) During mounting of swing back tooling =  6.7 V.

3) Electrometer was moved 5mm away from the ERM to allow for space during swingback = 6.4 V.

4) During swingback went up to peak of = 25.2 V.

5) Once Electrometer was moved back to being 1" from optic (and in chamber team in home position) = 24.9 V.

6) At this pont we broke for lunch. Once everyone out the chamber and soft door cover on = 23.1 V.

7) When we came back from lunch (1 hour later) = 13.7 V.

8) When the soft door cover came off = 12.2 V (with slow decay).

Note: - We had a first round failed discharge procedure. Panel was not flapping in gap. Decided to stop and start again. More details to follow.

9) After completion of ETM / ERM Gap discharge procedure (2nd time) = 3.8 V (and decaying). This was done at 2:56 pm.

10) After swing back of ERM (back to original position) and with electrometer reset to being 1" from optic = 9.3 V  (and decaying). This was at around 3:04 pm.

Note: - At this point team started to suspend the chains and run transfer functions on both the main chain and the reaction chain. We left the electrometer in place during this time.

11) At end (prior to doors on) at 4:12 pm = 4.2 V *  (and decaying).

12) Final number - At the end we put the cap back on the electrometer and it was now reading 1.3 V. Therefore we are calling the final number = 2.9 V (i.e. 4.2V - 1.3V).

Note:-  The humidity got as high as 41% during efforts. At end before doors on it was 32%.

Note: - Doors were finally in place at around 5 pm.

Kate Gushwa, Gary Traylor, Danny Sellers, Travis Sadecki, Betsy Weaver and Calum Torrie

Summary

Last weekend I used the offset-lock methof to measure the cavity length noise of the OMC.  I tried this with a single-bounce beam last fall, but the intensity noise from the IMC was too large and it contaminated the signal.  Here, with the carrier light from the full IFO, the intensity noise is low enough that a good measure of the displacement noise can be made, at least below 100Hz.  Here I describe how to offset-lock the OMC and how the cavity length noise is measured; also I make a projection of the OMC cavity noise contribution to DARM.  Noise due to OMC length fluctuations should be >10x below the DARM noise floor, except around the ISI resonances at ~1kHz.

This result is complementary to the cavity length noise measurement that Koji & I made by PDH locking the OMC with a single-bounce beam.  In comparison, this half-fringe (or "offset-locking") method is limited by intensity noise, but it's less invasive and can be performed entirely from the control room.  It might be useful for, e.g., estimating the OMC cavity noise during PEM injections or remeasuring the displacement noise after we install dampeners on the HAM6 ISI (still speculative). 

[Yet another method is estimating the displacement noise from the dither-locking error signal; Koji has done this in April and it agrees with the PDH result below 10Hz, but above 10Hz the dither error signal is limited by sensing noise.]

Offset locking the OMC

Here are the steps to lock the OMC on the half-fringe, while the full IFO is locked (and the DARM offset engaged).  This is essentially the same technique as I used last fall (and is halfway coded in the OMC guardian), which was itself a copy of what Zach did at LLO.

• First, sweep the OMC to find the carrier TEM00 mode.  Lock the OMC on the carrier with the OMC-LSC dither and record the average power.  For the measurements detailed below, the power into the IFO was 2.3W, the DARM offset was 3.7e-5 counts (~42pm), and the DCPD Sum was 20.43mA.  Engage the DARM boost (FM7) to reduce intensity noise.  The OMC should be aligned with the QPD loops for the entirety of the measurement.
• Unlock the OMC, turn off the input to the length servo, turn off the boost and integrator, and turn off the dither to PZT1.  Zero the OMC-LSC_SERVO gain.

The next steps adjust the OMC-LSC error signal path to set the offset locking at the half fringe point:

1. Turn off the input to DCPD_NORM_FILT, and set DCPD_NORM_FILT_OFFSET to be equal to the DCPD_SUM when the OMC is locked on the resonance (in my case, 20.43mA).  Now when the OMC is locked on the half-fringe the input to the dither demodulation should be 0.5.
2. Turn off the highpass (FM1) and bandpass (FM5) fitlers on the input to the demodulation (this is the OMC-LSC_PD_IN filter module).
3. Set the dither frequency to 0 Hz (normally it's 4100 Hz), and set the demod phase to 90deg (normally 0deg).  This will demdulate the DCPD_NORM signal at DC and pass the cosine term to the length servo.
4. Set the OMC-LSC_SERVO_OFFSET to be 1/2 the gain of the cosine demodulation, that is, OMC-LSC_SERVO_OFFSET = 0.5 * OMC-LSC_OSC_COSGAIN.
5. Turn on the length servo, turn on the integrator, and turn on the gain.  I wound up using a very small gain, 0.01 gave a UGF of ~8Hz.  You may have to try this a couple of times, since there is a sign ambiguity that depends on which side of the fringe you're on.
6. Once you're locked, check that DCPD_SUM is 1/2 of what you measured on the full-fringe lock.  I had to adjust the length offset (step 4) to get true half-fringe locking.  Not sure why.
7. Measure the intensity noise on the OMC transmitted light (you can use DCPD-NORM_OUT_DQ), take an OLTF so you can correct for the loop suppression, and you're done.  Use the SDF table to undo all your filter changes.

Measurement of OMC Cavity Length Noise

Figure 1 illustrates how the offset locking technique uses the RIN on transmission to measure the cavity length noise (delta-x).  There's a theorist on my thesis committee, this is for them.

Figure 2 is the transfer function of the length servo, while locked on the half-fringe.  This was used to correct for the loop suppression.

Figure 3: the top panel is a plot of the measured RIN during the half-fringe lock (red), and the same data corrected for the loop suppression (blue).  The RIN has a noise floor of ~1e-7 / rt[Hz] above 100Hz.  This noise is coherent between the DCPDs and is well above the shot noise + dark noise level.  (To demonstrate this I plot the NULL stream, along with the DCPD dark noise; both of these were convereted to RIN-equivalent noise by dividing by the DCPD_SUM during the offset lock, 10.2mA.  The shot-noise RIN for 10.2mA is in good agreement with the NULL channel.)

I think the noise floor is due to intensity noise coming from the IFO, and above 100Hz we are not measuring RIN due to cavity length fluctuations (except around 1kHz, more on this later).  This 1e-7/rt[Hz] noise floor matches the noise floor of the ISS 2nd loop PDs (not plotted), but the coherence is not =1 everywhere.  Why isn't it filtered by the IFO?

Figure 4 is the cavity length noise derived from the RIN measurement, red is the in-loop data, blue is the 'free-running' length noise (corrected for loop suppression).  The displacement noise is calculated from the RIN using the RIN/dx relationship, derived from the cavity transmission formula.  For the OMC, locked halfway up the fringe, the conversion is 2.72e-9 meters/RIN - to get this number you need to know the transmissivity of the OMC input and output mirrors, it is 8000ppm.

Figure 5 is a comparison between our measurements of the OMC displacement noise: Zach's measurement from L1 (red); Koji's result from the PDH lock of the OMC (black); and this measurement (blue).  This plot is an update of LHO:16174.  In general the measurements are in agreement.  The PDH result has a lower noise floor above 100Hz; probably the other measurements are limited by intensity noise.  The half-fringe result is lower than the others around 10Hz; Koji measured the noise from the PZT drivers in January, this may be the limiting noise source for the current measurement.  (Zach says that the L1 measurement is possibly limited by laser frequency noise in the 3-40Hz band.)  The peaky excess in the half-fringe result around 1Hz and 3Hz are coherent with intensity noise and are not real cavity motion, see Fig. 3.

Figure 6 is a zoom of the region around 1kHz.

Projection to DARM_ERR

Here I estimate the contribution of OMC cavity length noise to the DARM noise.  Valera, Ryan, and Sam discussed this in G1100903, and I think my methods are essentially the same as what they describe.

First, the noise in the light transmitted through the OMC due to the (suppressed) cavity length noise can be calculated two ways:

1. by modeling the quadratic coupling of length noise --> DCPD_SUM when locked on the full resonance
2. by modeling a bilinear coupling between the RMS cavity offset and the length noise.

For the quadratic case, we calculate the noise in the power measured by the DCPDs due to the cavity length noise as:

dP[f] = (4*F/lambda)^2 * ASD[x(t)**2]

...where F is the cavity finesse (=391), lambda is 1064nm, and the final term is the ASD of the square of the cavity length noise, x(t).  This is a quadratic approximation to the peak of the Fabry-Perot transmission curve, for the OMC it is accurate for motion less than ~3e-11 meters, which is true for our data across all frequencies.  [Important: you have to take the ASD of the square of the time-series, not the square of the ASD -- the Fourier transform is a linear operator but squaring is not.  Thanks Sheila!]

Alternatively, the bilinear coupling is estimated as:

dP[f] = (4*F/lambda)^2 * 2 * x_rms * ASD[x(t)]

I've used x_rms = 3e-13 meters, although this could be an overestimate (see comment below).

Next, to convert dP[f] into an estimate for the DARM noise budget, I turn the crank on the DCPD_SUM --> OMC-READOUT_ERR calculation that is described here:

OMC-READOUT_ERR = DCPD_SUM * (Pref / P0) * xf**2 / (2*x0) * G

...where Pref = 1.56 (power normalization), P0 = 24.1 (input power, Watts), xf = 14 (fringe offset, picometers), x0 = 15.8 (DARM offset, picometers), and G = 8.56e-7 (overall gain factor).  These are the values during the lock on Jun 7 at 00:00 UTC that Evan used for his latest noise budget.  This calculation converts DCPD_SUM in mA into counts at the DARM_IN1 error point.

Finally, to convert counts of DARM_IN1 into meters of CAL-DELTAL displacement noise, I use a factor of 5e-7, a zero at 389Hz and a pole at 7kHz.  The factor of 5e-7 was estimated by matching DARM_IN1 to CAL-DELTAL_EXTERNAL_DQ in DTT.  This is about 3x larger than what I expected, based on the calibration of the DARM offset in counts to picometers: 1e-5 counts / 14 picometers.  Not sure why there is this discrepancy.

In Figure 7 I plot the CAL-DELTAL spectrum from June 7 with estimates of the contribution from the OMC, using the quadratic and bilinear coupling results.  As mentioned before, the OMC length noise projection above 10Hz is probably limited by intensity noise and is bogus, except for the peaks around 1kHz, which are about a factor of two below the noise floor.  If this result is correct, it's an ongoing mystery as to why we don't see this 1kHz noise in the DARM spectrum.  Robert has investigated the acoustic coupling in this band and also projects that some noise at ~900Hz should be peaking above the noise floor.  Why don't we see it?

For interested NoiseBudgeters, the attached text files have the data for the red (quadratic) and blue (bilinear) traces in Fig. 7.  (You shouldn't think this noise source has gone unattended in the noise budget until now: Evan says he had measured the OMC-->DARM noise coupling some time ago, using excitations into the OMC-LSC loop, and had convinced himself the OMC wasn't a source of broadband noise.  No reason to think any different from this study.)

Summary: A seismometer was buried 40 m from EY to take advantage of strong attenuation of the tilt signal relative to the acceleration signal from distant sources. Seismometers located outside the buildings may be useful in reducing problems associated with tilt.

Our buildings are tilted by wind. Our seismometers do not discriminate between this tilt and acceleration (like a pendulum), so wind-induced tilt produces spurious control signals that can make it difficult to lock and maintain lock. A tilt sensor can be used to discriminate between tilt and acceleration, but we may also be able to discriminate between the two by taking advantage of source differences in the band below 0.5 Hz, where tilt generates the largest spurious acceleration signals. The tilt in this band is mainly generated locally by the wind pressure against the walls, while the acceleration signal is mainly generated by ocean waves and other distant sources.

While the seismometers are in the far field for most low frequency accelerations, they are in the near field for building-generated tilt (wavelengths at 0.1 Hz are about 50 km). To take advantage of the rapid attenuation with distance in the near-field, I buried an STS-2 seismometer in a meter deep hole I dug 40 m from EY (Figure 1).  Figures 2 and 3 show a comparison of the buried seismometer to the SEI seismometer on the ground inside the EY station. During the low wind period the spectra look virtually identical below 0.5 Hz and the coherence is high because the sources are mostly distant and the wavelengths are large. During the high wind period, the buried seismometer doesn’t change that much (there is some change in the microseismic peak because high/low wind were about 24 hours apart), but the building seismometer shows a huge signal from tilt.  The coherence in windy conditions is low because the tilt is highly attenuated 40 m from the building. Figure 4 shows spectra for higher wind, 15-35 MPH.

Of course the external seismometer would not detect real building accelerations due to the wind. But if the unwanted tilt signal dominates over the wanted wind acceleration signal, a seismometer outside the building may be useful, and I think that this is the case below about 0.5 Hz.

I note that I first tried this in the beam tube enclosure tens of meters from EY, but found that the wind tilts the beam tube enclosure almost as much as the building (but not coherently), supporting a hypothesis that aspect ratio is the important variable. Also, Jeff points out that, if we insulate the buried seismometer well (and I did not) we might even be able to do better than the building seismometers with their current insulation during even low-wind periods. Figure 5 is like Figure 2 except that the signal from the stage 1 Trilliums is included.

I set up an accelerometer on the beam tube and measured the accelerations as I simulated the tapping that I observed during cleaning, and my tapping experiment mentioned here: https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=19038. Examination of the time series indicated that accelerations during typical taps reached about 1 m/s^2.  I suspect that some taps reached 1g. The figure shows spectra for metal vs. fist taps. The above link notes that I was unable to produce glitches while hitting the beam tube with my fist. The fist and metal taps both had about the same maximum acceleration of 1 m/s^2 (ambient acceleration levels are about 4 orders of magnitude lower), so the source of the difference in liberating the particles is probably not acceleration. Instead, it may be the change in curvature of the beam tube, which would be expected to be greater at higher frequencies. The responses from metal taps in figure 1 peak at about 1000 Hz while the fist taps peak at about 100 Hz. These results are consistent with a hypothesis that the metal oxide particles are liberated by fracture associated with changes in curvature rather than simple vertical acceleration.

When I was producing glitches in DARM for the link above, I noticed that we did not get glitches every metal tap but every several metal taps. I tried to quantify this by making many individual taps at several locations along the beam tube. Unfortunately, we lost lock with the first loud glitch and I did not get a chance to repeat this before the vent. Nevertheless, it would be good for DetChar to look for smaller glitches at the time of the taps given below. Allow a 1 second window centered around the times given below for my tap timing uncertainty. I tapped once every ten seconds, starting at ten seconds after the top of the minute so that there were six taps per minute.

UTC times, all on June 14

Location Y-1-8

20:37 - 20:38 every ten seconds

20:47 - 20:50 every ten seconds

Next to EY

20:55 - every ten seconds until loss of lock at 20:56:10

We had four known reasons for having difficulty locking today, one is an unsolved mystery that might be hurting us more often than we realize.

• The BS limiter was off durring DRMI lock acquisition, this is turned on and off by guardian.  I've attached a screenshot of the gaurdian log.  The code should have returned true imediately after having turned off the limiter, and proceeded onto the next state, however, it ran the LOCK_DRMI_1F run again.  The requested state at this time was DRMI_1F_OFFLOADED, which means that this guardian should have moved onto the next state.  Is there a way to search the Guardian logs to see if we have had other instances of this?
  • we have seen this behavoir in the past, and Jamie suggested that this could have been because the state LOCK_DRMI_1F was requested.  I know that was not the issue today, but I changed the state so that it is no longer a requestable state anyway, since it should never be requested.
• The roll mode damping for ETMY was switched off at the output.  All the roll mode damping settings are now enforced by the guardian, using ezca.get_LIGOFilter().only_on.  

I reloaded both ISC_DRMI and ISC_LOCK guardians today, to incorporate these changes.  

• We had difficulty transitioning to ETMY ESD, evan is writting an alog about what we think is the solution
• Now the winds are quite high, the forecast says they will die down around 1 am.  

Good luck TJ!