ITMX HWS adjustments (More picoing and 800nm bandpass added)

Dan B. , Danny V.  

• The first time powering up to 30 W the  ITMX Hartmann beam profile seemed resonable and showed a semi-circular profile.
• Next two lock sequences, the ITMX HWS was outputting nonsense. We were convinced that this could be attributed to the new ifo alignment but after heavy picoing we didn't have much luck.
• We checked the ITMX Hartmann beam with both Hartmann plate and 1064 nm notch 800 nm bandpass filter removed and adjusted the picomotors to co-align stray 1064 light and ALS to the SLED beam. (please see the images 3-5)
• After witnessing some fringing in the Hartmann images, which we have previously attributed to stray 1064 light passing through an on-table beamsplitter (alog 45022,alog 45074), we suspected that the current 1064 notch filter 800 nm bandpass in place may be leaking some IR light to the ITMX Hartmann sensor. In response, we placed the 1064 filter as well as the 800nm bandpass filter from ITMY HWS 1064 filter from the ITMY HWS path in the ITMX HWS beam path in series with the 800 nm bandpass. The last attached image shows the state of the ITMX Hartmann sensor after these changes. 
• I'm not convinced that the result is much better than the first 30 W power up of the day.
• Investigations on what happened to the SLED beams after this first power up are pending. 

POWER UP attempt - more luck in nominal lownoise

- To power to 30W, we turned off the 'BOOSTS' (FM5 in DHARD P&Y and FM3 in CHARD P, labeled 'boost10W').

- Turned ISS 2nd loop off and back on with AC coupling (ISS_ON).

- Stepped up to 30W nominal  lownoise configuration.

- Stepped up the RP compensation gains to 1.6 at 30W (same sign as ASC control)

- We lost it on a simple SRC ASC yaw runaway (most likely SRC1 Y).

- We got a good range reading (with good calibration lines) at 25W - 85Mpc.

First look at REFL A and B demod phase and spectrum

After Fil and I swapped the REFL_B cable (alog 45652) I connected LSC-EXTRA_AO_2_EXC to IFO REFL CM board EXC and injected frequency noise in ENGAGE_DC_VIOLIN (2W) to see the demod phase of REFL_B and REFL_A.

First attachment shows the Q/I ratio obtained from the frequency noise scan. Top right shows the demod phase "error" obtained by p=atan(Q/I). Apparently REFL_A demod phase is decent, REFL_B is not terrible but could be improved.

Demod phase error should be flat but it isn't, it seems to cross zero at around 100Hz for REFL_A (maybe the demod phase was set using 100Hz excitation?). HOM or something else, regardless of the cause, this seems to mean that if we optimize the demod phase at f>300Hz there's some I-Q mixing in f<80Hz or so, and vice versa. Since I don't see any real Q signal for f>300Hz but there seems to be some junk for f<100Hz (second attachment), we might be better off optimizing at low frequency.

Not too much low frequency coherence between REFL_A and B, it's not clear if this comes from demod phase difference. Since we're not using REFL_B for control, from this point on we could rotate B phase digitally.

It's also worth measuring this at high power.

daytime work today, new ASC instabilities

Jenne, Hang, Sheila, Stefan

• This morning Jenne and Hang tried to use the alignment offsets set for 30Watts at 2W, and saw that the recycling gain was ~38 with those offsets at 2W.
• We then tried to revert the offsets to what we had before the vent, but saw the CSOFT instability when we started to increase the power.
• We have continued to have difficulties with the OMC alignment controls saturating.  This is not likely to be due to the HAM6 HPI being locked, because it is only 10-20 urad away from it's unlocked position, and because the beam is centered on AS_A and AS_B using OMs. 
• People were very enthusiastic about increasing the power, so we agreed to increase the power today and spend tomorrow after maintence on low noise locking at 20-25W. 
• We have been loosing lock due to an instability which seems similar to what is described here: 43968  Jenne and Stefan widened the notches mentioned in that alog, but the problem is still occurring at slightly lower frequency.
• We have had instabilities when powering up due to the SRC pitch loops.  This is new today, and is happening even when powering up to 20W, which had not been a problem before. 

violin mode changes

ITMY mode 8 used to be damped using - gain and no phase shift, now it is damped with positive gain and no phase shift. 

ITMX mode 9 used to be damped using no phase and + gain, now it is no phase and -gain.  These changes are in gaurdian.  

All of the changes that we have had to violin mode phases are on the ITMs, which makes me wonder if it is possible that changing the ring heater settings could change something about the violins.  

DC centering 8Hz notches were found on - should be off

Not sure the last time someone tried to run an ASC sensing matrix measurement, but it looks like the script didn't properly turn off the 8Hz notches in the DC centering loops.  I've turned them off.

DTT somehow swaps measurement test points

I was measuring H1:LSC-REFL_A_RF9 and H1:LSC-REFL_B_RF9 test points this morning (attached left).

It didn't make sense and I went back looking at the corresponding DQ channels at the exact same time (attached right).

As you see, if you believe that the DQ channels are correct, REFL_A were somehow swapped with REFL_B in test points, or vice versa.

I cannot reproduce it now, DQ channels and testpoints do agree with each other now, and they do look similar to DQ channels from this morning, which makes more sense (because REFL_B channels are calibrated in volts/W Volts, REFL_A counts/W counts, so the former should be smaller).

TCS Chiller Water Level Top-Off

No water was added, both chillers were overfilled/at max

Shift Summary - Day

TITLE: 12/01 Day Shift: 16:00-00:00 UTC (08:00-16:00 PST), all times posted in UTC

STATE of H1: Commissioning

INCOMING OPERATOR: None

SHIFT SUMMARY:

16:00 (8:00) Start of shift

17:28 (9:28) Karen to EY

17:34 (9:34) Kyle to LVEA -- check on turbo pumps

17:44 (9:44) Kyle out of LVEA

17:45 (9:45) Fil to LVEA -- check illuminator cover on HAM1

18:09 (10:08) Vanessa to EX

18:36 (10:36) Karen leaving EY

18:39 (10:39) Betsy, Travis to MY, grab 3IFO pre-mode cleaner

18:44 (10:44) Gerardo to LVEA - valve in NEG1, NEG2

18:58 (10:58) Gerardo out of LVEA

19:20 (11:20) Betsy, Travis back from MY

19:56 (11:56) Kyle to LVEA - VAC work

20:07 (12:07) Kyle out of LVEA

20:49 (12:49) Karen to MY

21:10 (13:10) Karen leaving MY

21:22 (13:22) Mike, guests to End Stations - tour

21:26 (13:26) Bubba to MY

22:42 (14:42) Travis, Betsy to LVEA -- drop off 3IFO equipment

22:44 (14:44) Jason to LVEA -- grab equipment from oplev cabinet

22:51 (14:51) Jason out of LVEA

22:55 (14:55) Travis, Betsy out of LVEA

23:09 (15:09) Christina, tow truck picking up GSA car

00:00 (16:00) End of shift

PCAL MASTER Front-end Model Bug Fixes

S. Karki, J. Kissel

As sudarshan pointed out last week (see LHO aLOG 45587), there were some bugs / inconsistencies between the recent front-end implementation of the explicit calculation of the PCAL optical efficiencies and force coefficients (calculations described in the appendix of T1800046; implemented in LHO aLOG 44418).

I've fixed the model (new screen shots attached), and have restarted the front-end code for both h1calex and h1caley, both of which use the library part I changed, 
    /opt/rtcds/userapps/release/cal/common/models/PCAL_MASTER.mdl
which has been committed to the repo.

The updated screen is also attached, and committed to
    /opt/rtcds/userapps/release/cal/common/medm/PCAL_END_FORCE_COEFF.adl

WHAM1 Table Top L4Cs Analog High/Low Gain evaluation

Hugh w/ many thanks to JeffK for super patience schooling.

With the new L4Cs inside the HAM1 Optics Table and the option to run the the interface chassis with an extra 10x gain available, the signals were cast into ADC digitization point voltages and compared to the ADC noise and its saturation level.

Bottom Line: *) These sensor will run with the analog gain in the high state.  *) V1 signal does not look like V2 & V3 below 100mHz.  *) The H1 signal does not look like the vertical sensors, naturally; but, I'm not sure it looks right.

Details:

First plot is the IN1s with LOW analog gain during quiet ground motion.

Calculation is: IN1.cts x ADCv/cts / 44[podpreamp] / 2[sensor chassis gain] / 275v/m/s[L4C cal] / 2pif x zpk(pair(1,45),[0 0],(2pi) ^-1).  This should put the counts to meters.

This looks reasonable when compared to the L4C Noise but happy if anyone looks closer at the details.

The second plot shows the H1 signal (just because it had the largest signal) while the sensor interface was in low gain cast back to volts at the ADC.  Periods of quiet ground motion and during a 6.0 EQ in Columbia are shown.  This is not too big an earthquake for us at LHO with the 30-100mHz BLRMS hitting just 1500nm/sec.  Also on the plot are the L4C noise curve when in low gain, the ADC Noise, and the ADC saturation level comparable to the DTT effective noise BW.  Note where the sensor's curve dances with the ADC Noise curve.

The third plot shows similar to above except the sensor interface is in high gain and the earthquake is the very large 7.0 in Alaska pushing the LHO ground motion to 30000 nm/sec in the low band.

So, conclusion: In low gain, the L4C is limited by the ADC noise at low and high frequencies; and, in high gain, there is still 2 orders of magnitude of headroom before saturation even during very large ground motion.  Conclusion, run in high gain.

These data and the matlab script resides in /ligo/home/hugh.radkins/Matlab/HAM1_TT

 

H1 PCALX Temperature Sensors Broken

J Kissel, N. Lecoeuche

While glancing at the PCAL END station overview screens, we noticed that the PCALX screen words "STATUS" and "Pcal modules are not powered" were blinking rapidly. The channels that this status indicator are measuring are:
    "Pcal modules are not powered"
        H1:CAL-PCALX_LASERENABLE 
        H1:CAL-PCALX_LASERPOWERCONTROL 
        H1:CAL-PCALX_SHUTTERSTATUS  
        H1:CAL-PCALX_LASERDIODECURRENT
    "STATUS"
        H1:CAL-PCALX_TRANSMITTERMODULETEMPERATURE 
        H1:CAL-PCALX_RECEIVERMODULETEMPERATURE

While the "Pcal modules are not powered" channels are flat and unchanging -- and the same as PCALY -- we found that the "STATUS" channels, i.e. the temperature sensor channels, were reporting a wildly oscillating temperature from 0 to 20 deg Celsius, with a ~1 sec period. 

This junk has apparently been happening since Sunday Nov 11 2018 (~a month ago) at 23:38:56 UTC (Nov 11 2018 15:38:56 PST).

Attached are screenshots of when the failure happened, and how the sensor outputs currently look.

Incoming updates to BSC-ISI and SEIPROC models

Tomorrow there will be a big update to the BSC-ISI and SEIPROC models. This is mostly to get code to smoothly ramp between different sensor correction filters and get the ground common mode signals into the ISIs. The former should make it smoother to switch between different seismic configurations. The latter is an attempt to improve isolation and reduce drives during earthquakes, something seismic group has talked about for a while.

Part of this update will include a simplification to how the ground sensor calibration and distribution is done. Currently all of the corner ISIs receive the 3 STS2s from a distribution chassis, and each chamber applies the same calibrations on all 9 channels. The update will use the ITMY ISI to send the uncalibrated signals to the SEIPROC model, which will do the calibration, then send that calibrated signal back out the ISIs. SEIPROC will also take over the BLRMS channel calculations.

I'm documenting what I had to do for the corner station ISIs here.

First screen shot, for ITMY I removed and the top level ground block which did the calculation of the ISI-GND_STS_HAM2/HAM5/ITMY calibrated STS channels. This block got moved to SEIPROC.

Second screenshot is the update to ITMY ground in. The green STS tag connect the adc inputs through a bus to IPC channels, of the form H1:ISI-ITMY_A_GND_X_IPC_PCIE, these go out to the SEIPROC model. The calibrated IPCs come in to the STS_CAL tag, the common mode signals come in on the CM_GND tag. Current all of the BSC models are set up to just use the ITMY STS (connected to the STS_CAL tag), but it might be a good idea to use some version of the current STS sensing matrix so that we can change which STS we use for feed forward on the fly by changing an epics variable. Both the STS_CAL tag and CM_GND go into the master model to the  h1isiitmy ITMY ST1 SENSCOR DOF block, contents shown in the third screen shot. This uses the code outlined in T1800414. There is also a new isi master part, the inputs are mostly the same, but there are fewer outputs because there is now need to ouptut the STS signals on the top level.

Fourth screenshot shows the uncalibrated IPC channels coming into the SEIPROC model, fifth shows the new, topnamed ISI block for the generation of the calibrated ground signals. This preserves the ISI-GND_STS channel names, so should not affect anyone who uses these channels. The signals are calibrated using mostly the same stuff that the ITMY model used, and include the BLRMS calculations, shown in the sixth screenshot. I will probably have to make a new MEDM to access the STS calibrations, though.

This update also applies to the endstations, but the STS calibration is done in a block added to the top level. I added this block a couple weeks ago, which required deleting a block like ITMY and stuff on the top level for the BRS subtraction, then wiring up a new block that organized it all in a library part. This change is shown in the seventh screenshot. The endstation ISIs are also getting the common mode signals from the corner.

New IMC drive align gains

After running a2l_min_IMC.py, the new IMC drive align gains are:

MC1-P2L:  -0.212     MC1-Y2L: -0.589

MC2-P2L: +0.485     MC2-Y2L: -0.572

MC3-P2L:  -0.345     MC3-Y2L: -0.594

Accepted differences in SDF for MC3. Screenshot attached.

RGA data before & after valving in two corner NEG pumps

RGA text file attached with scans before-during-after valving in last two NEG pumps in corner station (NEGs within several feet of RGA). Gerardo will note the times that he valved in each pump in order to correspond to RGA scans. We were looking for changes in H2 partial pressure but didn't see much change by simply eye-balling the RGA scan; attached is PDF of scan after pumps valved in - H2 partial pressure was at ~ 1e-7 A before and after.

3rd OMC preparations

[Betsy, Koji] @Bonding lab

The OMC is sit in the bonding lab for curing of the epoxy. Otherwise, it is ready to be moved to the chamber side.

- 3rd OMC optics cleaning

Attachment 1:
We applied FirstContact cleaning of the optical surfaces. The optical side of the breadboard was wiped with IPA-soaked cloth (without touchting the optics).
The FCs will be left until the OMC is brought to the chamber side.

- PD replacement

Attachment 2:
The original OMC DCPDs were replaced with the new high QE DCPDs. The PDs at the BS transmission and reflection sides are from the PD cage A slot 3 (A3) and A4, respectively. They correspodns to the PD serials B1-01 and B1-16. The final testing at Caltech showed the QEs of 0.980 and 0.981 respectively.

The FirstContact seals are attached on the PD apertures to prevent particulates come into the PD surfaces.

- Mounting blacket bonding reinforcement

Attachment 3:
We added a glass prism to reinforce the bonding of one of the mounting blackets on the top (suspension) side of the OMC breadboard. A small amount of glue residue form the existing bond was removed by a razor blade to clean the place for the new prism. As small amount of glue as possible was applied to have round glue foot print, particularly on the glass-glass joint. The glue on the Invar-glass joint looks round. The glue on the glass-glass joint looks square inspite of our effort (We really don't think it is an issue). A steel block was added to hold the prism until the epoxy is cured.

 

 

The test data for B1-01 (DCPDA, that in transmission of the OMC TRANS BS) and B1-16 (DCPDB, that in reflection of the OMC TRANS BS) can be found here in the 40m eLOG 255:
    https://nodus.ligo.caltech.edu:8081/OMC_Lab/255

The serial numbers are indicated in the OMC DCPD wiring chain,
    D1300502