WP 5941 The computer controlling remote access for CDS was rebooted to fix a BIOS setting.
FRS 5725 WP 5940 Dataviewer has been updated to version 3.0.4 to properly display double precision trend data for Ubuntu 12, Ubuntu 14, Scientific Linux 6, and Gentoo systems. WP 5946 The NDS2 client software software has been updated to nds2-client-0.12.2 to provide better error messages on failure to deliver data for Ubuntu 12, Ubuntu 14, and Scientific Linux 6.
Sheila, Rana
We saw the dIzumi/dBallmer instability happening again tonight at the 19->23 W transition. Since this was happening in the cSoft DoF, we started debugging it and made some changes to see if we can damp the instability with a more rational loop than using oplevs or ISS.
Balancing the input Matrix:
We drove lines in cSoft and dSoft and then adjusted the relative gains for the X & Y TransMons so that the crosscoupling was < 15% (as measured at ~8 Hz). Confusingly, we only adjusted the matrix elements for dSoft. The existing cSoft matrix elements already gave a good decoupling. Before the adjustment we saw (via step response) that big DC steps in cSoft made steps in dSoft. After the matrix change this behavior is gone. The step responses seem well decoupled.
After much struggle, we were able to measure the soft loop shape. After turning up the gain from -0.05 to -1.14, we have a ~30 sec time constant and the UGF is a little below 100 mHz. We are able to see that the mismatch between the radiation pressure modified soft mode resonance and the digital compensator (FM10) is what makes cSoft go unstable if we turn up the gain further. Tomorrow we make a better filter and then see if we can damp the angular instability with an angular feedback loop.
Because we used up all the filter banks for the fundamentals, I made broadband filters for the high ETM 2nd harmonics.
ETMX 1003.664, 1003.778, 1004.078 and 1004.537 are damped with MODE9 FM1, FM4, FM10 (notch at 1003.907 Hz) with -50 gain.
butter("BandPass",4,1003.67,1004.54)gain(120,"dB")*gain(100,"dB")* ellip("BandStop",2,3,40,1003.807,1004.007)
This filter was later found to be ringing 1003.778 Hz. It has to be reconfigured. The individual damp settings for 1003.664, 1003.778, 1003.907, 1004.078, and 1004.537 are 0deg, +-180deg, 0deg, +-180deg, and 0 deg.
ETMY 1009.44 and 1009.49 are damped with MODE10 FM1, FM4 with +150 gain
butter("BandPass",4,1009.44,1009.49)zpk([0;0],[],1,"n")*gain(100,"dB")
ETMY 1008.45, 1008.49, and 1009.02 are damped with -60deg, -60deg, and 0deg (all + gain). No broadband filter has been made yet.
The other high amplitude modes that I haven't tried to damped are ETMX 1005.17, 1005.94, and 1006.5
ETMX MODE9 FM1 now covers 1003.667, 1003.778, and 1003.907. The face are correct here and should be damped with positive gain setting.
ETMY MODE 9 FM1 now covers 1008.45, 1008.49, 1009.02. The phase is close to what damped last night. Still has to be tested.
[Jenne, Sheila]
We have had dither lines on for the BS and SRM, both pitch and yaw, for the last hour or two while we've been trying to increase the laser power. This is meant to be reminiscent of the test by Stefan and Elli back in August (alog 20777 and alog 20827), where we can look at how the demod phase in the AS36 signals is changing as a function of laser power.
Sheila may have some transfer function points that she looked at, at different powers, but a more full analysis will come during the daytime.
Lines were put in using the ADS of the ASC.
BS pit 30cts at 5.7Hz
BS yaw 30ct at 6.4Hz
SRM pit 100ct at 6.2Hz
SRM yaw 1,000ct at 6.8Hz
Shift started out with Jenne & Rana with H1 making Oplev OLG measurements. After they handed H1 over, I tried tweaking the arm alignment, but Y-arm was too far out, so ran an Initial aligment. Rest of the night has been focused on locking and high power investigations by Jenne/Sheila.
As noted by Gerardo's alog, CP5 will be alarming after a step change from 92% down to 84%. It's been holding steady here (but have been getting short MAJOR alarms and then longer MINOR alarms); if level drops below 74%, call the Vacuum Team.
Alignment Notes:
(John, Gerardo)
CP5 level has been changed from 92% to 84%. This is going to generate alarms since 84% is the low level alarm, please do not ignore alarms below 84%.
CP5 is now in PID control.
I think Gerardo meant to say you can ignore alarms near 84% but if the pump level falls much below that - say 74%; then action should be taken.
If this is the case you may set the liquid level control valve (LLCV) to manual and use the current manual setting of 90%. Or call one of us.
Vacuum staff will have alarm texts sent to their cell phones if the level drops below 80% (or above 99%)
Corey, Jenne, Sheila
We had an epics freeze while in the increase power state, and lost lock when epics started to update again. This happened at 2016-06-21 02:10:27
While relocking after this we had a guardian ezca connection error connecting to H1:ASC-OUTMATRIX_Y_2_3. We were able to connect to the channel using caput, but the error didn't clear until we took the ISC_DRMI guardin to stop and then back to exec.
Later in we had at least two more freezes (based on the ASC strip tools) around 6:50 UTC and 7:10:48 UTC.
CP2 needle valve shaft completely decoupled from actuator. The packing nut was too tight. Reconnected and set the packing nut a little looser. We should consider using a locking gel on this joint.
Peter, Stefan
We read up on alog # 20699, indicating that the main problem during power-up is a change of the MICH ASC error signal. That alog suggested mixing in A36I, but back then AS_A_RF36 was phased differently.
Thus we looked at the MICH_P signal in AS_A_RF36, and indeed confirmed that the signal is now in Q, as expected.
We thus set the following pitch matrix element for MICH:
ISC_library.asc_intrix_pit['MICH', 'AS_B_RF36_Q'] = 1 # reduce from 2 to 1
ISC_library.asc_intrix_pit['MICH', 'AS_A_RF36_Q'] = 0.666
With this setting we noticed that MICH_Y was unhappy - in particular the AS_A_RF36_Q_YAW signal grew tremendously during power-up. We confirmed my moving the BS that the same error signal combination as for MICH_P also made sense for MICH_Y, i.e. we used the following yaw matrix element for MICH:
ISC_library.asc_intrix_yaw['MICH', 'AS_B_RF36_Q'] = 1 # reduce from 2 to 1
ISC_library.asc_intrix_yaw['MICH', 'AS_A_RF36_Q'] = 0.666
These settings are now in the Guardian (including setting them to 0 in the DOWN state).
For reference, for SRCL we are using the following error signals:
ezca['ASC-INMATRIX_P_6_3']=0 # off for now # AS_A_RF36_I
ezca['ASC-INMATRIX_P_6_7']=1 # good enough for now # AS_B_RF36_I
ezca['ASC-INMATRIX_Y_6_3']=-1.5 # AS_A_RF36_I
ezca['ASC-INMATRIX_Y_6_7']=1 # AS_B_RF36_I
We then noticed that the SRC1_Y top stage offloading resulted in a slow oscillation, indicating the the loop gain dropped enough to make the offloading oscillate. We have not fixed this in the Guardian yet.
We also used the TCS_ITMX_CO2_PWR and TCS_ITMY_CO2_PWR guardians to set the TCS power at 11W. This ramps the central heating down as the power comes up. This is currently only done in the guardian after the whole power-up is finished (i.e. in the COIL_DRIVER). For now, we did this by hand.
With this setting we were stably sitting at 11W until the 5.4 magnitude earthquate in Vanuatu hit.
=================================================================================
After the quake had settled down, we went to 23W, and noticed that SRC1_Y now would prefer a different error signal:
ASA36_I to SRC1_Y: -1.5 (was -1.5)
ASB36_I to SRC1_Y: 0.53 (was 1.0) # (not in guardian yet)
We noticed that for large SRC1_Y offsets AS36_I seems to turn around - during one lock we even were able to keep it locked with opposite gain, al;though the sideband buildups were worse.
We also set
H1:TCS-ITMX_CO2_LSRPWR_MTR_OUTPUT = 0.2 # (not in guardian yet)
H1:TCS-ITMY_CO2_LSRPWR_MTR_OUTPUT = 0.0 # (not in guardian yet)
this slightly improved the sideband buildups.
We left it locked atr 25W to damp the violin modes out over night. Attached are screen shots of the ASC input matrix.
An example of how to follow along at home, and how you can access the matrix elements a little cleaner using the guardian interactive shell:
servo:~ 0$ remote_epics -u jameson.rollins lho
IFO = H1
ifo = h1
SITE = LHO
site = lho
GATEWAY = lhoepics.ligo-wa.caltech.edu
jameson.rollins@lhoepics.ligo-wa.caltech.edu's password:
Connection to lhoepics.ligo-wa.caltech.edu established
Searching for a free port to use for EPICS CA transport
The script will randomly select some ports in an attempt to find
an available network port to send the EPICS data over.
Attempting to use port 27963 for EPICS
Forward attempt 1
Launching bash shell setup to access EPICS at LHO.
Use the 'exit' command to return to your regular environment.
servo:~ 0$ guardian -i ISC_LOCK
--------------------
aLIGO Guardian Shell
--------------------
ezca prefix: H1:
system: ISC_LOCK (/home/jrollins/ligo/src/userapps/isc/h1/guardian/ISC_LOCK.py)
MANAGER SYSTEM:
nodes = NodeManager(['ALS_COMM', 'OMC_LOCK', 'LASER_PWR', 'ISC_DRMI', 'TCS_ITMX_CO2_PWR', 'TCS_ITMY_CO2_PWR', 'ALS_DIFF', 'ALIGN_IFO', 'ALS_YARM', 'SEI_BS', 'IMC_LOCK', 'ALS_XARM'])
Type 'nodes.init()' to initialize.
In [1]: ISC_library.asc_intrix_pit['SRC1', 'AS_A_RF36_I']
Out[1]: 0.0
In [2]: ISC_library.asc_intrix_pit['SRC1', 'AS_B_RF36_I']
Out[2]: 1.0
DHARD measurements were taken in alog 27832 with 10w laser power, and I compared them with the modeling results here.
We found that the measurements of Yaw fit better to the modeling than the pitch, which is also true for 2w situation (27617).
I removed one of two SR560s used for ISS 3rd loop and reshuffled connections so we can directly go to the 2nd loop error point.
First attachment shows the new configuration, second attachment the old one.
PD58 path is used as the injection point of the 3rd loop.
Since we're injecting to a different point, you need to figure out a new gain/filter.
Kiwamu says that you cannot increase power while the 2nd loop is on. The setpoint voltage for the second loop needs to be changed as the power goes up, otherwise the loop hits the rail, but there's a pair of 0.1Hz hardware poles in the offset voltage path and you cannot update the voltage fast enough. You might want to put digital double zero-poles, e.g. zpk([0.1, 0.1], [5, 5], 1, 'n') or something in H1:PSL-ISS_SECONDLOOP_REF_SIGNAL_ANA_CALI to mitigate this effect.
Anyway, this might mean that you need to use different gain/filter configurations for the 3rd loop while powering up without the 2nd loop, and once you reach the final power you enable the 2nd loop and switch to the final configuration for the 3rd loop.
Possible switch combinations:
PD select = 1-4 | PD select = 5-8 | |
ADD SUM5-8 = ON |
2nd loop ON, |
2nd loop OFF, 3rd loop ON with twice the gain |
ADD SUM5-8 = OFF |
2nd loop ON |
2nd loop OFF 3rd loop ON |
We're not losing the out of loop sensor, as PD5-8 analog summation output is NOT used as such.
Use H1:PSL-ISS_SECONDLOOP_SUM58_AC_OUTPUT and H1:PSL-ISS_SECONDLOOP_SUM58_REL_OUTPUT instead, these are digital sum and digital RIN.
Jenne, Rana
In order to symmetrize the hard/soft drives, we want the ITM and ETM actuation functions to be identical (i.e. the PUM pit to TM pit TF). Previously Evan and Rana adjusted the TOP stage damping to make all the TMs the same. We also balanced the common / differential drives using the AS port WFS.
Today we realized that this was not as good as we thought since the PIT optical lever servos were on for the ITMs, but no other OL servos were on. So the PIT DoF was an admixture of hard/soft all of this time. So we measured the OLG for all of the TMs and turned on the loops for the ETMs, so now the PIT OLDAMPs are on for all the TMs. The same filters are ON for all 4 loops, and we adjusted the gains to make the OLG sweeps look the same (to within ~10-20%):
G_ITMX = -300
G_ITMY = -300
G_ETMX = -440
G_ETMY = -300
Attached is a measurement of the pitch oplev loops for each test mass, after we've adjusted the gains so that they match. It's unclear why ETMY's phase is a little different.
It worked s expected! Adding the OL damping on the ETMS symmetrized the drives and made the hard loop TFs more smooth - less notches. Only Pitch though; yaw TF still looks weird.
Conor, Jeff K Summary: - Conversion of ISI Suspoint motion into the DoF basis works for the arms, with limited fidelity. - Corner station DoFs seem not to work, possibly some problem with inclusion of the Beamsplitter. - IPC communication between ISI and SUS at ETMY is spoilt somehow. The GS-13s on stage 2 are used as witnesses for residual motion. Their outputs are calibrated to nm, and converted to suspension point motion using Euler matrices (Cart2Eul). The output is in the local suspension coordinates. For example, the channel 'H1:ISI-ITMX_SUSPOINT_ITMX_EUL_L_DQ' is the longitudinal motion of ITMX, which is along the normal vector pointing outward from the HR surface, in the global +X direction. These suspension point motions are IPC'd to the OAF model, and in the case of the ETMs this involves some multiplexing/AI-filtering for communication down the arm. In OAF they are combined into the IFO degrees of freedom based on T1500610 . As a sanity check, I fetched the ISI-model Suspoint channels and matrix'd them into the IFO DoFs in Matlab, for comparison with the OAF DoF channels. For the arms (XARM, YARM, CARM, DARM), things seem to be somewhat okay. I confirmed this by eye, and then took the spectrum of the subtraction (attachment 1, CARM_Spectrum). The residual is consistent with the dynamic range of single-precision floats (about 10^9), at least at high frequencies. A quick software-only transfer function using four poles at 0.1Hz and white-noise injection in a spare filter bank shows a similar total dynamic range (attachment 2, DTT_dynamic_range). The anti-imaging filters, for transmission of the ETM signals along the arm, should allow a considerably smaller residual than we see at low frequencies, and I don't really know what else might cause this. The corner station DoFs are substantially wrong, visible directly in the time series' (eg attachment 3, MICH_time_series). Since MICH only uses the ITMs (confirmed to be 'okay') and the Beamsplitter, presumably the difference arises there. PRCL and SRCL are similarly incorrect. Document T1500610 has some minor errors in the Suspoint->IFO-basis definitions, but the current matrix has correct values as far as I could see. It also points to the Suspension model versions of the Suspoint motion, eg 'H1:SUS-ITMX_M0_ISIWIT_L_DQ', but the model uses the ISI versions. Initial testing revealed discrepancies between: 'H1:ISI-ETMY_SUSPOINT_ETMY_EUL_L_DQ' 'H1:SUS-ETMY_M0_ISIWIT_L_DQ' visible in attachment 4 (ETMY_IPC_issue). The other 3 test masses didn't show this gross level of disparity.
The problem with the corner station calculations are a model mapping error. I pointed this out to Jeff but he must have thought I fixed it or he just spaced it out given his usual very full to-do list. I've never edited the OAF model and did not feel comfortable doing so. Attached is the errant part of the model. The BS PRM PR2 and PR3 are all mis-mapped in this section.
Dave, Conor, Jenne, Hugh
Jenne is fixing the mapping mis-wire for the corner station calculations.
Dave checked the data path and found the SUS and ISI versions of the CART2EUL matrix for the ETMY were not the same. This looks likely to be the issue rather than an IPC problem.
I checked the values against the data file I used to populate the matrices for all the suspensions and it still agrees. I suspect the values in the SUS ETMY matrix are out of date.
See alog 11036 for more details: Data file is
/opt/rtcds/userapps/release/isc/common/projections/ISI2SUS_projection_file.mat
Here is some detail on the GS13 signal paths. The six GS13 signals coming out of the ISI model are split, with one set going via Dolphin IPC to the SUS system, and the other set going through the ETMn_CAL into the ETMx_SUSPOINT part, in which is passes through a CART2EUL matrix. The SUS model receives the six dolphin channels, passes these through the ISIINF part, then through its CART2EUL matrix into the ISIWIT. It is these two CART2EUL matrices which have identical settings at ETMX and are slightly different at ETMY.
J. Kissel I've compiled, installed, and restarted the OAF model with the SUSPOINT to IFO basis transformation bug fix, and committed the top level model, /opt/rtcds/userapps/trunk/isc/h1/models/h1oaf.mdl Attached is a screen shot showing the current and correct channel ordering.
J. Kissel I've also updated the SUS version of the ETMY CART2EUL matrix (i.e. H1:SUS-ETMY_M0_CART2EUL channels) with the values found in /opt/rtcds/userapps/release/isc/common/projections/ISI2SUS_projection_file.mat that was apparently discrepant in only *some* of the rotational terms, and only by ~10%. I can't explain why these were off, and I'd trended the values for the past 850 days (with the stop time of June 2015, so I covered the time install time mentioned in LHO aLOG 11036) and they've not been changed. One of life's mysteries, I suppose. Fixed now! For the record, the new values are | L | | 0 -1.0000 0.2000 0 -0.2823 0 | | X | | T | | 1.0000 0 0.4294 0 0 -0.2823 | | Y | | V | = | 0 0 0 1.0000 -0.4294 0.2000 | | RZ | | R | = | 0 0 0 0 0 -1.0000 | | Z | | P | | 0 0 0 0 1.0000 0 | | RX | | Y | | 0 0 1.0000 0 0 0 | | RY |
Also, I posted some analysis of the relative motion between HAM2 and HAM3 in the DCC https://dcc.ligo.org/G1601390 differential Z is really small. differential X kind of sucks at the microseism. Not sure how much it matters for the IMC, but it is worth thinking about. I think there is some fruitful work to be done here - if it would be useful to reduce the relative motion of the ISI tables in the corner at the microseism.