Pulling First Contact from ETM-X Replace door on BSC9 BSC10 on roughing pump today. Plan to start turbo pumping on Monday BSC9 on roughing pump after door is on. Plan to start turbo pumping on Monday Electrical cabling work on PEM and OpLevs in the LVEA
Since we turned on cleanrooms in each end station the temperatures have drifted several degrees - ie. out of control. In addition our outdoor temperatures have warmed significantly.
Our electric heat in the outbuildings can be turned on in 15kw stages - which is also roughly the heat output of one large cleanroom.
Last night I decided to turn off the heaters in both buildings to help restore the temperatures to the setpoints of 68F and 65F for EY and EX respectively. Once we turn off the cleanrooms we will consider turning the heat back on. We should attempt to coordinate this so please notify Bubba or John prior to shutting off cleanrooms.
Age related failure of Pirani channels reading the same value -> MKS 937B will be substituted from this point onward via attrition
I left a measurement running overnight -- this will sweep the 9MHz sideband frequency by 20kHz or so and inject a 1kHz oscillation into IMC_F.
To turn off the measurement, select MOD DISABLE on the IFR source in the highbay, and turn off EXC A on the IMC Common Mode Servo screen.
I was trying to figure out how to run the MC alignment offset calibration this afternoon, but ran into some difficulties with new features of the guardian / IMC ASC system. I wasn't able to turn off the ASC loops without the IMC alignment drifting. This wasn't an issue previously, though that was back in February.
Unfortunately, after tinkering with some of the ASC switches etc., the IMC alignment seems to have got into a bad state where it locks, and then drifts in alignment, and then unlocks after about a minute. I put the guardian mode back to exec, and requested locked, but each time the IMC locked the alignment drifted until lock was lost.
To avoid a possible runaway drift overnight, I've put the guardian requesting down.
I came back late to the lab and found the same problem. In order to take some measurements I set the IMC WFS gain to zero. The IMC was stable after that.
J. Kissel, P. Fulda, K. Izumi, K. Venkateswara While trying to get MICH locked again for seismic diagnoses of goodness, Kiwamu found that the H1 SUS ITMY's highest vertical (bounce) mode at 9.7 [Hz] had been rung up. It's the end of the day and no one wants to stick around long enough to really diagnose why it rung up, or attempt to damp it. Our best guess is that the mode got rung up while the SEI team was measuring HEPI tilt-decoupling coefficients (see attached 12 hour trends). Paul and Mackenzie might keep trying, but will most likely just wait until tomorrow for the mode to ring down naturally.
J. Kissel Stefan needed measurements of the HLTS M3 stage for global angular controls (see LHO aLOG 15727), and after taking the first set I found / re-remembered that Betsy and Evan only half finished the import of LLO's damping loop design for the HLTS (see LHO aLOG 15329, and subsequent comments). As such, I've installed the complete set of filters and gains LLO HLTS damping in both H1 SUS PR3 and SR3. I attach a screen shot of how both SR3 and PR3 damping filters are configured (2014-12-18_H1SUS?R3_DAMP.png). M3 to M3 transfer functions with these new damping loops on (shown in the measurement vs. model comparison, allhltss_2014-12-18_Phase3b_H1HLTSs_M3_Don_ALL_ZOOMED_TFs.pdf) is what Stefan and Elli used for the design of the PR3 WFS / Oplev control. Such that design sticks, I've - Foton File - Removed all old filters - Committed new filter file to userapps repo - Safe.snap / EPICs configuration - Turned on the gain-only filters (in FM7 of every DOF), and set all of the EPICs GAINs to -1, as is the case for other suspension types (instead of having the overall gain of the loop in the EPICs GAIN field, as is done at LLO -- see LHO aLOG 15364 for documentation). - Turned off some M3 DRIVEALIGN elements that have been accidentally left ON - Captured a new safe.snap - Committed new safe.snap to the userapps repo. A model of the LLO filters now exists, and lives in the following design script: /ligo/svncommon/SusSVN/sus/trunk/HLTS/Common/FilterDesign/Scripts/design_damping_HLTS_20141218_LLO.m stored .mat file of filters: /ligo/svncommon/SusSVN/sus/trunk/HLTS/Common/FilterDesign/MatFiles/dampingfilters_HLTS_2014-12-18.mat model: /ligo/svncommon/SusSVN/sus/trunk/HLTS/Common/FilterDesign/MatFiles/dampingfilters_HLTS_2014-12-18_model.mat I've attached the model figures of merit, including the loop design, global control TFs, sensor noise, actuator noise, residual seismic noise, and a total assessment of the performance (though I was lazy and didn't add the impulse response that I'd added for the BSFM, QUAD, and HSTS; Next time). Again, I don't both making a comparison with the previous filters because they weren't at all designed with any thought put into them and did not meet any kinda of desirable noise performance, have desirable global control transfer function response, nor did they have a low-Q / low-ring-down-time. The transfer function templates live here: PR3/SAGM3/Data/2014-12-18_2331_H1SUSPR3_M3_WhiteNoise_L_0p01to50Hz.xml PR3/SAGM3/Data/2014-12-18_2331_H1SUSPR3_M3_WhiteNoise_P_0p01to50Hz.xml PR3/SAGM3/Data/2014-12-18_2331_H1SUSPR3_M3_WhiteNoise_Y_0p01to50Hz.xml SR3/SAGM3/Data/2014-12-18_2328_H1SUSSR3_M3_WhiteNoise_L_0p01to50Hz.xml SR3/SAGM3/Data/2014-12-18_2328_H1SUSSR3_M3_WhiteNoise_P_0p01to50Hz.xml SR3/SAGM3/Data/2014-12-18_2328_H1SUSSR3_M3_WhiteNoise_Y_0p01to50Hz.xml I've committed new versions of /ligo/svncommon/SusSVN/sus/trunk/HLTS/Common/MatlabTools/ plotHLTS_dtttfs_M3.m plotallhlts_tfs_M3.m to the SusSVN because I've updated them to include the ability to compare against a damped model and I've added the two new measurements.
Here are the Tilt Decoupling results for ITMX. The coupling was smaller than ITMY reported earlier. See the left side plots for the inline(Pitch) tilt. The blue is before and the brown trace is after but not much improvement. I didn't get a chance to calulate and try the crossline(roll) tilt coupling but the data is there and I'll try it soon.
See the second attachment for a shot of the IPS Align Matrix.
Elli, Thomas, Stefan After Jeff redid the PR3 M3 P2P and Y2Y transfer function measurements, we designed actuation filters for high bandwidth feed-back to PR3. We decided to use M3 on PR3, offloaded to M1 (top). The filters we used are: PIT PR3^-1: zpk([0.1+i*0.82;0.1-i*0.82;0.5+i*3.4;0.5-i*3.4],[0.5+i*3.02;0.5-i*3.02;11.1111+i*38.4258;11.1111-i*38.4258],1,"n") YAW PR3^-1: zpk([0.2+i*1.23;0.2-i*1.23;0.3+i*2.95;0.3-i*2.95],[0.3+i*2.5;0.3-i*2.5;11.1111+i*38.4258;11.1111-i*38.4258],1,"n") In both filters the design philosophy was to compensate the mechanical poles with a slightly lower Q digital zero,but such that we never run out of phase for unconditional stability. The resulting transfer function works as resonant gain, suppressing the motion on resonance. Attached are loop transfer functions, both measured (red) and calculated based on Jeff's measurement times the applied filter.
The ITMx HWS is ready to take measurements. I have left it off for now.
The ITMy HWS SLED is clipping the periscope before entering HAM4. I will adjust it when we have green light from ETMx again.
Jim, Hugh, Krishna, Fabrice:
we have been chasing problems in HAM3:
- one aspect of the problem is that the Z sensor correction doesn't perform well. A large peak at 0.6 Hz appears in the spectra, in all directions, when the sensor correction is ON.
- another aspect of the problem, related, is that there is low coherence from the ground seismometer to HAM 3 geophones. This is more a broadband effect.
- another puzzling effect is that the Z sensor correction makes X, Y performance better.
We have been investigating to find out whether this is a CPS sensor noise issue, a mechanical issue (rubbing, hitting, mechanical shortcut....), and/or a loop shaping issue.
Some comments and results of the investigation:
- page 1, the sensor correction is off, the ISI is damped, the coherence from ground to HAM3 is excellent.
- page 2, isolation loops and sensor correction are ON. The coherence drops at 0.6 Hz.
- page 3, the sensor correction is off. Coherence between ground to ISI is no good. But the coherence between ground and HEPI is good. So HEPI is likely not introducing the noise.
- page 5, I ran harmonic tests to see if something is hitting. We did not find anything obvious.
- page 6, I compare all units, in damping mode. HAM 3 looks fine.
- page 7, I turn ON the isolation on all units. HAM 3 still looks fine.
- page 8, I turn ON sensor correction on all units. The peak appears on HAM3.
- page 9, I turn off the Z sensor corretcion on HAM3. The peak is gone, but the X motion is much higher...
- page 10, I also turn ogg the Z sensor correction on HAM2, to verify it has no effect in the X direction for this "good" unit.
- page 11, I put HAM 2 and 3 vertical loops in high blend (Z, RX,RY), and turn off the sensor correction. HAM3 is fine.
- page 12, I turn on the sensor correction (X,Y,Z), HAM3 shows a slight peak at 1.2 Hz.
- page 13, I put Z back in low blend (01_28), sensor correction is still ON, HAM3 is still fine.
- page 14, I put RX and RY in low blend (250mHz), HAM 3 looks fine. So the problem shows up only with certain combinations of sensor correction and blend filters.
- page 15, I put RX and RY back in the standard 01_28 blend configuration, the peak at 0.6Hz is back...
- page 16, I repeat the test in page 14 (using 250 mHZ filters on RX and RY), the peak is gone again....
So at that stage I started to check loop shapes.
- page 17, I checked the blends installed. All seem fine. 0.6 Hz turns aout to be frequency where the CPS complementary filter crosses unity, but I think it's just a coincidence.
- page 18 to 22, I checked the open loops. Rx seemed to have a drop near 0.6 Hz. But the measurement was not repeatable, as shown in page 23 which looks fine.
- page 25, 26, I check the super-sensors. They don'y look nice at 0.5 Hz, but it may not be related.
Did not find anything abvious in the loop shapes, so back to sensor noise hunting:
- page 27, HAM 3 in damping mode. No problem in the local sensors.
- page 28, back to standard config (01_28 blends, and sensor correction...) , we can see the peak at 0.6 Hz in most local sensors.
- page 29, I put RX and RY of HAM3 in high blend, the peak now shows up at 1.12 Hz.
Summary:
the problem really depends on which blend filters are engaged for RX and RY. Depending on the blend filter used, the peak can show up at .6 Hz (using 01_28), 0.12Hz (high blend), or very little (using 250 mHz). THis set of blend filters however is too agressive at low frequencies. Despite the measurements we have performed today, it is still difficult to assess wheter this a a CPS issue, a mechanical issue, or a loop issue. We keep investigating.
Fabrice Krishna Hugh.
Krishna was suspecting that RX tilting on ITMY and the BS was impacting the HEPI Z Sensor correction results. Sure enough, when checked it was most coupled on the BS Z to RX and next on ITMY HEPI Z to RX. The other couplings, that is, HEPI Z to RY, and for ITMY both HEPI Z to RX & RY, where less by about a factor of 10.
The Measurement
HEPI Z is driven with a 0.001 to .1 band pass excitation (see attachment 1) looking for coherence with ISI Stage1 T240 X & Y.
The HEPI is in normal configuration, Lvl1 position loops but with sensor correction off.
The ISI Stage1 all dofs are put into high blend (T750) and its sensor correction is also off.
Once a baseline of the existing HEPI Z to ISI Stage1 T240 to RX & RY coupling is established as seen by the T240 Y & X, the HEPI is then Tilted in RX & RY with a smaller magnitude but similar bandpass to see the actual tilt of the ISI Stage1 when HEPI is tilted. The Decoupling factor is computed by the ratio of the former to the latter: RXz/RXrx. This correction goes into the IPS Align matrix.
Results
See attachment 2 for ITMY. The left three plots are the TF data for inline tilt on ITMY, this is the Y direction caused by RX; the three right plots are for the crossline tilt RY showing on X. The first step is shown in blue: the area below 0.1 to 0.01hz with good coherence is our tilt coupling. Notice on the right side, there is poor coherence and the TF magnitude is 10x smaller than the tilt in the Ydirection seen on the left side.
The green traces are the direct tilt measurement HEPI RX(RY) to ISI RX(RY). Picking a few magnitude points from the blue & green traces in the area of interest and averaging the ratios gives the decoupling factor blue/green= 0.23/46(e.g.) == -0.0049 with the sign coming from the phase which are ~180 out of phase.
The brown trace (only on the left side) shows the reduction with the decoupling factor in the matrix (see last attachment) when the first measurement is repeated. (I failed to save the coherence for this but it was reduced just above 0.8 from the near 1 at the start (blue.) This indicates there may be more improvement to be made but it will be time consuming and may not be worth it. The improvement though is clear, about a factor of 10.
Time constraints (commissioners) prevented a brown results curve measurement for the Z to RY tilting but I have the data to calculate the ratio. We expect the improvement to be minimal as the coupling is already low.
I've attached similar data for the Beamsplitter HEPI. The RX and RY correction values were based on the following measurement:
the transfer function between Z drive to X/Y
Correction value = ------------------------------------------------------------------------
the transfer function between RX/RY drive to X/Y
For the beamsplitter, we measured
RX correction = - 0.0172
RY correction = + 0.0038
The plot shows the transfer function between Z drive to X/Y before and after the tilt decoupling.
Rick and Dave:
We moved the MSR mac minis to EY and EX this afternoon to pair them up with their respective PCAL cameras via a local USB cable (procedure is https://dcc.ligo.org/LIGO-T1400755). The mac minis were returned to the MSR and reinstalled. The Y arm computer can ping the Y arm camera but image software does not connect. Rick will see if the camera/UT-1 are still trying to connect when he goes to EY tomorrow.
(Filiberto, Gerardo)
Installed two viewport protector assemblies on BSC9, ports G2 and G3. Took the opportunity to inspect both viewports, they look good.
Installation included mounting an illuminator on port G3 and a camera housing on G2.
The camera is not connected, will revisit tomorrow.
Kyle, Gerardo, Bubba -> ~0930 hrs. local -> Removed BSC9 West door Kyle -> ~1350 hrs. local -> Started pumping BSC10 annulus Kyle -> ~1500 hrs. local -> Decoupled pump cart from HAM1/HAM2 annulus (Climbed on and around HAM1 and HAM2)
LVEA: Laser Hazard Observation Bit: Commissioning 07:15 Karen & Cris – Second cleaning at End-X 08:15 Jim – Testing BSC10 08:25 Cris & Karen – Back from End-X 08:33 Krishna – Going to End-X to shutdown BRS ahead of door removal 08:57 Kyle, Gerardo, & Bubba – End-X door removal 09:00 Gerardo – Installing viewport protectors at BSC10 08:58 Richard – End-X to verify Laser Safe and to test interlock 09:03 Mitch – In LVEA west bay 09:15 Richard – Transition End-X to laser safe 09:18 Cris & Karen – In LVEA for cleaning 09:45 Betsy & Travis – Going to End-X 09:49 Filiberto – End-Y taking pictures ESD in-air cables 10:08 Rick & Jason – Going to End-X 10:15 Filiberto – Going to End-X to drop off power supply 11:00 Gerardo – LVEA HAM1 & 6 to check vacuum comp 11:00 Doug – Working on HAM4 OpLev alignment 11:20 Gerardo – Out of LVEA 11:31 Vending machine delivery 11:37 Kyle – Back from End-X 13:15 Kyle – Pumping the BSC10 annulus at End-Y 13:30 Power cycle Video4 monitor 14:05 Kyle – Start pumping BSC10 chamber at End-X 14:15 Dave & Rick – Going to End-Y to work on cameras 14:43 Kiwamu – Cleaning up tools at HAM1 14:49 Kyle – Disconnecting pump cart from HAM1 15:02 Kyle – Out of LVEA 15:30 Dave & Rick – Back from End-Y
F. Matichard, K. Venkateswara
On Tuesday (12/16/2014), Fabrice and I got an opportunity to test the ETMX ISI in different configurations. The wind speed was in the 7-12 mph range at EX. The configurations I tested were:
A) ISI Damped
B) ISI isolated with High (T750) blends. Inertial isolation happens only at frequencies above ~750 mHz.
C) Nominal LLO blend configuration. Inertial isolation happens above ~45 mHz.
D) (C) + sensor correction using LLO filter.
E) (C) + sensor correction using Rich's Z sensor correction filter and the tilt-subtracted super-sensor.
The first page of the first pdf shows the ground seismometer (RED) and the tilt-subtracted super-sensor output (BLUE) in displacement units. It shows factors of ~5 subtraction below ~50 mHz. The second page shows the T240 on Stage 1 under the different configurations. The following page shows the CPS and the final page shows the T240 RY.
Looking at Page 2, Configuration E appears to improve over D, roughly by factors of 2-4 below 0.5 Hz, while showing no excess amplification at low frequencies, as measured by the CPS. This is in agreement with modeling predictions described here. Unfortunately, I couldn't make a measurement in (E) but with BRS off for comparison, because someone walked into EX VEA and drove up the BRS amplitude.
Measuring this improvement interferometrically has not been possible using the ALS (green laser system). It is likely that ALS is noise limited in this frequency range as described here. Measuring it using the IR laser needs much more dedicated time which is not easy to come by.
So far, we have not tried using different blends for the inertial sensors. Some improvement may be possible by blending higher in X on Stage 1. But so far, it looks like BRS can improve ISI performance by factors of 2-5 in the 50-500 mHz range. Improving performance at the microseism might significantly improve detector robustness as seen during ER6 at LLO.
The other two attachments show the coherence between various sensors when the ISI was damped and when ISI was in the (E) state.
Secretary notes from BSC9 today:
We will pull the FC sheet ~9:30 am tomorrow and hopefully get the door back on by lunch.
Note, there were apparently no witness plate or witness optics placed on the floor of BSC9 when it was closed last time, so we did not pick them up today.
Also, I forgot to mention that when we inspected the ETMx-HR, we saw some mottling and haze across the surface but we did NOT see the "ring" feature that the FC left on the ETMy-HR.
[Alexa, Mackenzie, Paul]
This morning with some free time on the IMC, we were going to re-run the alignment offset calibration procedure in preparation for a beam jitter measurement (a la aLOG 9870 and aLOG 10016). However, I found that the script failed when looking for the channel H1:IMC-WFS_SWTCH. The IMC_WFS_MASTER medm screen still has a button that should be controlling this switch.
Alexa dug around in the IMC_WFS_MASTER.adl file to find out what that switch was linked to and found that it actually calls the scripts "/opt/rtcds/userapps/release/ioo/h1/scripts/imc/mcwfson" and "wfsoff".
Looking into the mcwfson script, we saw that the script sets the WFS gain to 0.25 (writing to H1:IMC-WFS_GAIN), and attempts to switch the missing channel H1:IMC-WFS_SWTCH to 1.
Running the script from the terminal gives the error: channel H1:IMC-WFS_SWTCH not accessible. A quick caget gives the same result.
Does anyone know where this channel went?
This switch was removed in favour of another trigger switch at LLO. I believe the related alog to this change is this LLO alog 10933 by David F