I re-arranged the Beckhoff SDF part of the CDS Overview MEDM to add the missing PLC4
The previous ROC was only showing rate-of-change for the unity time periods (1MIN, 1HOUR, 1DAY). For the other periods it was just showing the value differences.
My new system (rolled out yesterday) has three EPICS channels per time period: R, D and V.
V is the value acquired n-minutes ago. E.g for the 3HOUR period, it is the value acquired 180 minutes ago.
D is the value difference between the current value and V
R is the rate of change (D/period) in units appropriate to the period (either per minute, per hour or per day).
The second CP4 thermocouple was added. For performance reasons, the dewar channel was removed.
The bias slider values that were stored in the SDF SAFE file were old. When there was a boot fest last Thur of ETMY SUS (alog 40913), these values got restored and the ETMY no longer pointed to ~0 on the Oplev. Today I trended those values an dataviewer in order to restore them. The latest values which we mechanically pointed the ETMY to, used a little bit of bias:
PIT 34.4
YAW -105.0
However these values don't put it on the Oplev in PIT now. Today it wants bias slider values below in order to be well centered on the OL:
PIT -135.6
YAW -105.0
Whatever. So many mysteries, so little time. Door closed.
I have loaded these newest bias values into the ETMY SUS SDF SAFE file.
In light of the recent oplev issues (primarily the increased signal noise, which is an indication of laser glitching), I swapped the oplev laser for a spare so there is not an issue with the oplev during the upcoming ETMx work (as the oplev is being used as our alignment reference). The old laser is SN 130-1 and the new laser is SN 197-2. The new laser has not been tweaked for glitch-free operation, it will have to be replaced before the start of O3 with one that has.
Note, the current SUM on this OPLEV is now ~23,800.
ESD cables were tested with HIPOT tester. Cables tested to 1KV, all passed.
While working in the Chiller room, the PSL Chillers were serviced. 100ml water was added to the Crystal chiller. No water added to the Diode chiller. No changes in the filters for either chiller.
As I was crawling into bed......... I noticed that the temperature of the regenerating GN2 flow leaving the heater was 68C -> it should have been much higher. After 5 minutes, I refreshed the screen and it was then 55C -> I realized that the heating of the regeneration flow had stopped, sent Chandra R. and Gerardo M. a text and set off to the Y-mid. Upon arrival, I noted that the dewar vapor pressure was as expected (18 psig) and the flow of rengenerating GN2 was also as expected at 50 SCFH. In the VEA, I noted that the temperature of the GN2 leaving the heater was now 16C? 18C? Next, I noticed that one of the RED indicator lamps on the Regeneration Control Panel was illuminated -> I reset this before noting which of the two lamps was lit (they are side-by-side). Upon reseting, the GN2 temperature began to rise too quickly as the sepoint was 150C and the proportional gain was 7. -> I lowered the SETPOINT to a value slightly above the current value and also lowered the proportional gain value but the PID was too slow and the GN2 temperature continued to rise too quickly. Next, I turned off the heating at the control panel (small switch - not main Panel Power) and let the temperature stabilize. Once stable, I set the proportional gain to 1.00 and the SETPOINT value to 2C above the currently displayed value. Then, I turned on the Regeneration Heater. Now the "SHORTED SCR" lamp lit up - reset - goes off then back on etc... I can hear excess current (60 Hz hummmm) before trip. It is broken. I closed the regeneration and pressure build manual valves and shut down the Regeneration Control Panel. Also the 3-phase CB. So for tonight, only the air duct heater associated with the bake enclosure will be heating CP4. 2350 hrs. local -> Leaving site now.
J. Kissel
I've finally gotten around to creating a transfer function measurement script that can compare all OPOS transfer function results, assuming they've been exported and processed by
/ligo/svncommon/SusSVN/sus/trunk/OPOS/Common/MatlabTools/plotOPOS_dtttfs_M1.m
and that comparison script is called
/ligo/svncommon/SusSVN/sus/trunk/OPOS/Common/MatlabTools/plotallopos_tfs.m
With this new analysis infrastructure, I've exported the best L1 SUS OPO data from LHO aLOG 37508, which now lives in,
/ligo/svncommon/SusSVN/sus/trunk/OPOS/L1/OPO/SAGM1/Data
2018-01-29_0900_L1SUSOPO_M1_WhiteNoise_L_0p01to500Hz_tf.txt
2018-01-29_0900_L1SUSOPO_M1_WhiteNoise_P_0p01to500Hz_tf.txt
2018-01-29_0900_L1SUSOPO_M1_WhiteNoise_R_0p01to500Hz_tf.txt
2018-01-29_0900_L1SUSOPO_M1_WhiteNoise_T_0p01to500Hz_tf.txt
2018-01-29_0900_L1SUSOPO_M1_WhiteNoise_V_0p01to500Hz_tf.txt
2018-01-29_0900_L1SUSOPO_M1_WhiteNoise_Y_0p01to500Hz_tf.txt
and processed it with the above mentioned script. Sadly -- the LLO results are confusing on their own because they require a different calibration depending on whether DOFs are translation or rotation, and those calibrations don't correspond to any physical meaning as far as I can tell. I've merely scaled the DC gain of the TF to match the model (independently for translation and rotation, but the same number within each). This implies that the gain has something to do with the OSEM to Euler basis change -- and from what I've recently discovered with Alvaro, it might be a simple case of incorrect units in the rotation lever arm. I'll discuss this with Stuart and Alvaro to see if we can come up with a physical explanation.
So -- take LLO's data matching the model's [m/N] and [rad/N.m] with a deer-sized salt lick.
That being said, I'm then able to plot H1 and L1's results; see alloposs_2018-03-12_H1SUSOPO_Phase3a_ALL_ZOOMED_TFs.pdf.
Several things are immediately obvious:
- R2R and P2P are measured to be almost identical, but don't match the model very well.
- The L2L primary resonances are identical and match the model well.
- L1's V2V transfer function shows two primary resonances, where H1's only shows one. Both show cross-coupling to yaw, but the L1 results show *very* substantial cross-coupling to what appears to be the Yaw resonance, where as H1's only shows minor coupling.
- The nastiness we've been worried about in H1's Y2Y transfer function seems to match the L1 data, except for one cross-coupled pitch mode at 1.65 Hz.
I'm not sure if this tells us too much in the hunt for what's wrong with the H1 suspension, but it certainly provides a lot of good food for thought.
P.S. I spent a good bit of time this evening making sure everything committed to the SusSVN is now under the OPOS directory instead of the VOPO directory. This includes data from LLO. So please svn up your local copy ASAP, and begin filling out the directory structure that now conforms to the rest of the SUS repository.
Last Thur at 4pm local, the ETMX Oplev took another jump - this time ~70uRad in PITCH, negligable in YAW. At the same time, the PIT, YAW, and SUM signals became noisy and have been ever since. According to the THUR Mar 08 Ops summary alog 40914, the only crew was the in-chamber PCAL baffle crew, who were not around the OPLEV piers outside of the chamber and down the tube a bit. There apparently were no reboots. The hash on the signals is an apparent sign that the oplev laser is dying however. Jason reports that the PIT and YAW signals are normalized by the sum. There is a very small shift on the SUM signal at the jump on THUR but I'm not sure how it adds up to the PIT-only 70uR shift.
Due to this mystery and the PCAL misalignment (alog 40968), we launched Keita to check the ALS pointing at the table at EX. He found things somewhat well aligned using ~50uRad of bias on the ETMX/TMSX. So, at least the ALS beam and the hashy, twitchy OPLEV beam pointing monuments agree that the ETMX is still pointed to within ~50-100uR of the last good arm pointing (recall that the vent to air contributes to some pitch mis-pointing as well, see alog 40714). The PCAL reports a much larger magnitude of ~1mR of alignment error.
In the morning, Travis and Rick intend to remount the PCAL target on the front of the ETMX and see where the incoming PCAL beam is - if it's off, the whole periscope has been misaligned. If it's hitting the ETM at the right place, Travis will look closer at the ETMX itself. (Note, we have not yet touched the ETMX at all since the vent from last week - only the PCAL baffle crew took parts past it and into the manifold to work. The ISI has been locked however, but the PCAL was checked for good alignment before the start of the baffle work and after the ISI lock. alog 40892)
Also, Jason plans to swap the laser for a healthy one first thing tomorrow.
Hmmm... looking closer at the attached plot I see that at the end of the plot when we asked Corey to return the ETMX to zero mid-morning, he only had to use the ETMX PIT slider but I see a step in both the PIT and YAW trends of the OPLEV...
TBC... Anyone volunteer to sort this mystery?
Today, Hugh pointed out that the EX vea temp has been on the rise for the last few weeks due to all of the cleanrooms running. This isn't a huge surprise but it got us wondering if the steps in pointing of the ETMX as viewed by the OPLEV PIT are actually the suspension sagging and brushing something mechanical. (The steps account for ~100uR of unaccounted mispointing.) Indeed the top BOSEM sensors see the steps. However, a couple hours of staring at trends gets us no closer to this theory as the actually problem, for the following reasons:
1) Bringing the BSC9 volume up to air lifted the suspension, while the temperature increase is dropping the suspension, restoring it towards its original vertical position.
2) The temperature is hotter now than at the time of to step events (purple lines on the plot), yet transfer functions of ETMX V and P look healthy to me.
So, the agreed apon sentence from late yesterday still holds:
We will rezero the OPLEV now, with the biases from the arm lock - this should be good enough to find the arm again after pump down to within ~100uRad.
Recall, ALS pointing on the table yesterday with these slider settings looked pretty good.
Bias sliders:
PIT 7.3
YAW 40.2
As is typical for QUAD work, we had to iterate through a 2nd First Contact cleaning on the ETMY today which prevented us from closing out BSC10 today.
Recall, yesterday we applied First Contact to the ETMY HR surface (cone spray, with painted on reinforced edge. This morning we
- continued to clean the suspension and optics a little more (gingerly so as to not migrate particulate to areas we don't want it in (nooks and crannies around fibers, the back side of ETMY).
- we prepped the suspension for closeout via setting the top 4 ETMY BOSEMs to account for buoyance, and locked all of the nuts on ~60 plus barrel and face EQ stops (I could see particulate fall off of the silver plated nuts here and there as I did this with my tools and flashlight).
- blew the TMS lowest telescope large mirrors to clear them of particulate.
- we pulled the ACB back down into place and secured it.
- wiped a few spotty surface on the TMS, and we wiped the floor more.
- with Jim and the door crew on hand, we then pulled the First Contact. Things looked good with the pull until I inspected further and found 2 very small (sub mm x 1mm long) streaks of FC nearish the center of the optic. Darn. Abort.
We then called off the ISI unlocking/door crews, put the ACB back, reattached the cone and did another round of FC spray and paint application.
We'll go back at in in the morning.
I photographed the recently installed baffles on the EY P-Cal periscope. Figure 1 shows a before/after comparisons of views from the ETMY beam spot with labeled features. The glints in the “before” picture are thought to be responsible for the appearance of raven pecks in DARM (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=37630).
Figure 2 is a close-up of the reflections on the cylindrical portions of the Cryo and P-cal periscope baffles, suggesting that the light is from the gate valve and so will not be there when the valve is open. However, the gate valve reflections suggest a possibility: since the cryopump-baffle cylinder is not perfect (see reflection in Figure 2), and the cylinder may also be slightly angled, light from the ITM may reach the ETM through shallow-angle reflection off of the inner surface of the cylinder. I think that the ITM can "see" the inner surface so it should probably have been a cone or otherwise baffled.
Figure 3 shows a before/after comparison with much of the light from the gate valve blacked out. The remaining glints are from the finishing grooves on the view port blank-offs, where the lines of the groves are normal to the direction to the beam spot.
The P-Cal periscope baffles have not yet been installed at EX but I took photos to see if there are glints that won’t be covered. Finishing grooves that are deep enough and other linear structures could cause glints (see https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=40454 ) from regions of the periscope that won’t be covered by the new baffles. But Figure 4 shows that there are no such glints that will not be covered up by the current design.
We increased GN2 flow into CP4 reservoir with pressure build circuit (vaporizer underneath Dewar). Dewar head pressure is now around 18 psig and flow increased to 50 scfhx100 from 20. More flow equals more heat into the system. CP4 Dewar gets a truck delivery tomorrow so we'll make adjustments accordingly.
I plugged a couple holes on top of the enclosure and laid some left over rockwool bats and sheets of Al. Gil is shopping for insulation blankets that we can drape over the enclosure. MCE came out today to update the PLC software such that program doesn't HOLD temp. prematurely. We are considering adding 10 kW heater to achieve 130C.
VEA is getting borderline too warm. Average temp is 83.6F. We will turn chiller on before it rises above 85C for the safety of 3IFO and electronics equipment.
The bake temps have improved since today's mods (plugging a couple of holes). Supply is currently at 107C and bottom valve is 69C with heater output at 43%.
Terry Daniel
The TTFSS slow output is busted again. We discovered the same symptom with the temperature control input having a -14V offset. This seems to put the laser in a funky state: operating multi-mode would explain some of the other strange things we encountered before we identified the problem. At the end of the day, we were back were we left on Friday...
Chandra, Dave:
The upper alarm level for H0:VAC-MY_CP3_TE202A_DISCHARGE_TEMP_DEGC was increased from 130C to 150C.
The Rate-of-Change system was modified to calculate the ROC of H0:VAC-MY_CP4_TE253A_REGEN_TEMP_DEGC
The ROC for this TE253A was added to the alarms system, with the upper/lower alarm levels the same as TE202A (+1.0CperHour and -1.0CperHour).
ROC and alarms were restarted to make these changes.
J. Kissel, T. Shaffer I've finally gotten around to doing detailed processing of some of the new HAM Tip-Tilt Suspensions (HTTS), namely ZM1 and ZM2. ZM2 looks good, but not great -- the latest data set (from LHO aLOG 40254 back in January) shows hints of pitch cross-coupling into yaw -- but still totally acceptable for our intended use. ZM1 -- that which had a magnet polarity saga (identified in LHO aLOG 40847, solved in LHO aLOG 40851), and was after afterword rubbing -- still is rubbing, even after an unalogged attempt to fix it, but identifying that the UR flag looks poorly seated. I attach a newly revamped set of individual measurement HTTS plots, i.e. that produced by /ligo/svncommon/SusSVN/sus/trunk/HTTS/Common/MatlabTools/plotHTTS_dtttfs_M1.m (rev 9163) that correctly show the Euler Basis drive to OSEM sensor response plots. The EUL2OSEM plots (pgs 7-9) in the individual results .pdfs reveal an interesting difference between a functional suspension (ZM2) and ZM1 -- namely that the UL frequency response is markedly different from the model and from ZM2. Where the model expects (and ZM2 shows) a zero from both the UL and UR sensors at 1.3 Hz (ZM2 shows 1.03 and 1.11 for UR and UL, and only slight different overall response shape), ZM1 shows a poorly coherent zero at 1 Hz in UR, no zero in UL, and again a markedly different response from UR. Further corroborating evidence that ZM1's UL OSEM is problematic? - The individual OSEM sensor response to a simple, one-at-a-time, DC offset on the COILOUTF banks for ZM1, shows that UL exhibits less response than UR. - TJ says that the UL flag is poorly seated, and is thus coming in through the OSEM sensor / coil at an angle with respect to the nominal sensing / control axis. I also attach the usual comparison against other suspensions, using a new revision of /ligo/svncommon/SusSVN/sus/trunk/HTTS/Common/MatlabTools/plotallhtts_tfs_M1.m which now plots the phase in addition to the magnitude allowing us to #RespectThePhase This shows that ZM1's P resonances are slightly low in frequency, implying a heavier moment of inertia, and it's dreadfully obvious that the Y frequency is totally distorted into several peaks. The lower frequency at 1.17 Hz isn't even a cross-coupled L mode at 1.27 Hz. However, the upper-most frequency is indeed a cross-coupled pitch mode. TJ will investigate further once the opportunity presents itself later this week.
TITLE: 03/12 Day Shift: 15:00-23:00 UTC (08:00-16:00 PST), all times posted in UTC
STATE of H1: Planned Engineering
INCOMING OPERATOR: None
SHIFT SUMMARY:
Did not get to put a door on at EY. Needed to reapply a new coat of First Contact.
Note on ETMx Alignment:
PCal group mentioned issues with ETMx alignment earlier in the day. ETMx was in a DAMPED state and was pitched down and to the left. I took ETMx Guardian to ALIGNED and this took care of most of the yaw, but it was still pitched down. At this point I took the ETMx Pitch Alignment bias from:
51.6 to 96.6 urad.
Checking alignment of ALS to check ETMx (Keita). And SUS group is also investigating.
LOG:
NikoL, Rick S
We switched on the the Pcal laser to check the beam pointing after the vibration damper and baffles installation work last week.
We (with Corey's help) pointed the ETM to center the optical level signals.
With this alignment the Pcal spots are way off, more than 1 cm (see attached photo, both beams should be (and were earlier last week) centered on the integrating sphere aperture).
Keita is going to check on the TMS alignment so we can try to discriminate between the possibility that the Pcal periscope shifted during the alignment work and the possibility that the ETM pointing (or OptLev) has changed.
If the ETM pointing is off, it would be on order 10 mm over 10 m, or about 1 mrad.
TravisS, NikoL, RickS
It appears that the Pcal periscope structure moved (top pitched back, away from the ETM) during the "Baffles and Shields" installation.
The attached photos show the spot positions on the ETM target both before (last Wed.) and after (this morning) the installation of the vibration absorbers, Mt. Brackets, and barrel baffle panels.
Also attached is a detail of the Pcal ETM target (D1301014) showing the dimensions of the features on the target.
It appears that the upper beam moved up about 3.5 mm and the lower beam moved right about 3.5 mm.
The gaps between the centers of the upper flexures and the A7 adapter wall were slightly more than 0230". Our goal was to keep them below 0.220". We set them significantly below 0.220" last week (maybe 0.210") but apparently the additional mass of hte hardware added reduced the compression (as expected, see LIGO-T1800047-v2).
J. Kissel, M. Pirello Marc and I sat down with the mess of wiring diagrams today (see below) in order to verify that all cables had been connected and chassis have been turned on after the several month delay in re-arranging and installing the HAM5/6 suspension electronics due to vent activity. We found a few chassis not turned on, and a few cables in the wrong place, but those problems were easily rectifiable. I've now begun the verification process that all is well with that which are easy: the OM1, OM2, and OM3. I had to flip the sign of the damping loops from -1 to +1 (now captured in the safe.snap SDF file), but now all damping loops are functional. I've confirmed so by taking damped M1 to M1 transfer functions, which look clean and healthy, regardless that HAM6 is still at air and the ISI and HEPI are locked. I attach screenshots of the L to L transfer function, though I took all three DOFs. I'll process them fully in due time (likely once we're closing up HAM6). Next up -- ZM2 and OFI. %% Details %% Drawings used to identify cable names (which have three different naming schemes): For SUS SR3, SRM, and OMC: D1002740-v6 For OM1, OM2, and OM3: First Sheet of D1200666-v11 For ZM1, ZM2, VOPO, and OFI D1700384-v6 For ADC / DAC connections and assignments for all of the above D1002740-v7 Data files: /ligo/svncommon/SusSVN/sus/trunk/HTTS/H1/ OM1/SAGM1/Data/2018-01-19_0101_H1SUSOM1_M1_WhiteNoise_L_0p01to50Hz.xml OM1/SAGM1/Data/2018-01-19_0101_H1SUSOM1_M1_WhiteNoise_P_0p01to50Hz.xml OM1/SAGM1/Data/2018-01-19_0101_H1SUSOM1_M1_WhiteNoise_Y_0p01to50Hz.xml OM2/SAGM1/Data/2018-01-18_0102_H1SUSOM2_M1_WhiteNoise_L_0p01to50Hz.xml OM2/SAGM1/Data/2018-01-18_0102_H1SUSOM2_M1_WhiteNoise_P_0p01to50Hz.xml OM2/SAGM1/Data/2018-01-18_0102_H1SUSOM2_M1_WhiteNoise_Y_0p01to50Hz.xml OM3/SAGM1/Data/2018-01-19_0102_H1SUSOM3_M1_WhiteNoise_L_0p01to50Hz.xml OM3/SAGM1/Data/2018-01-19_0102_H1SUSOM3_M1_WhiteNoise_P_0p01to50Hz.xml OM3/SAGM1/Data/2018-01-19_0102_H1SUSOM3_M1_WhiteNoise_Y_0p01to50Hz.xml
Processed this data in detail finally. See attached. Note that this data was taken before the sign convention was identified as wrong for all HTTS in LHO aLOG 40847, so the phase does not agree with the model at DC by 180 deg. This has since been fixed, see LHO aLOG 40853 though I haven't yet retaken TFs that demonstrably confirm that all is well. We'll do so once we're closer to closing HAM6. Also note that the individual measurement results have been processed by a revamped /ligo/svncommon/SusSVN/sus/trunk/HTTS/Common/MatlabTools/plotHTTS_dtttfs_M1.m which plots the EUL drive to OSEM response correctly for the first time, and the plotall script, /ligo/svncommon/SusSVN/sus/trunk/HTTS/Common/MatlabTools/plotallhtts_tfs_M1.m respects the phase.