Transition Summary:
Title: 02/01/2016, Day Shift 16:00 – 00:00 (08:00 – 16:00) All times in UTC (PT)
State of H1: IFO unlocked – TCS commissioning work underway. Working on relocking.
Outgoing Operator: N/A
No restarts reported for all four days.
I remeasured the 45 MHz phase noise coupling into DARM, this time with broadband excitation. No surprises here; it looks the same as the swept-sine TF that we took before O1 (20783).
Setup is similar to the previous measurement, although in this case the PLL has a UGF around 20 kHz, so that the control signal going to the IFR can be read out directly in order to get the frequency deviation.
During this test, there was some glitching of the rfam stabilization.
The attachment shows the rfam OOL monitor (bottom) and the digital excitation to the IFR (top). Note the control signal glitches that appear even in the absence of IFR drive.
We have had multiple locklosses today from the beamsplitter coil driver switching.
This is puzzling, since this step was largely unproblematic during the run.
Opened FRS ticket 4325. Unfortunately for the this study, the only BS's coil driver monitor channels stored are noise and voltmons, both of which are upstream of the output impedance network (so they don't measure the current effect of switching the "acquire" network off as is done here) *and* they're only stored at 1024 at the fastest. I recommend we start by either (a) installing an analog voltage breakout pick-off in-line with the M2 BOSEM chain down-stream of the coil driver to identify the amplitude of glitching which takes out the IFO's lock, and address from there, or (b) Changing the h1susauxb123 front-end model to store the driver's FAST I MON at 16 [kHz]. (These can go in the commissioning frames, and also the noise and voltmon can be removed.)
Starting to look at this but have a question before I start. Why is the Ramp on UL 0sec and the other filter 10 Sec.
Richard refers to the ramp time in COILOUTF bank; however the ramping between coils is performed by the new Ramp Matrix part not this bank. It's likely that this COILOUTF bank ramp times were "set" some long time ago (clearly more than 300 days ago!) and because it's not used for any ramping of control signals, it has merely remained untouched.
This problem has gotten worse and better in the past without any known cause, for example durring ER7 it was particularly bad.
The changes to the common ASC_MASTER ASC library part have been made, and committed to the svn.
Late last week, I made the changes for the 2f centering loops. Today I added excitation points (and monitors) at the output of the ASC servos, which will make sensing matrix measuring easier. I also added 2 more DC centering loops, which will be used to feed the Trans QPD signals to the transmons. Finally, I created 2 filter banks CHARD_A and CHARD_B for each pitch and yaw, so that we can do some frequency dependent error signal blending.
All of these changes compile, but nothing has been installed or restarted. We'll do that on Tues.
Medm screens are mostly done. I still need to increase the input and output matrix sizes, but since all the new stuff is at the end of the matrices, even if we restarted the model right now, we could use the screens as-is. I also need to add Exc notifications to the ASC overview screen for the new sensing matrix excitations.
Today I've been working on the guardian controlled laser pzt servo. Using the Y-arm laser guardian can now scan the PZT, pick a locking point, calculate the slope at that point and use this to set the gain for the servo. The servo itself has a unity gain frequency of 0.5Hz and 1/f below that. After picking a locking point guardian sets the PZT to that voltage and waits for several seconds to allow the laser thermopile to come into equilibrium before re-measuring the power output which is then fed to the power setpoint. The servo is then engaged. This gets around problems that I was having where using the value taken directly from the power to voltage curve would cause a small but unmanagable offset in the setpoint at the point where the servo was engaged. The servo can cope with positive and negative slopes to the curve which reduces the number of times we will need to step the chiller temperature when trying to lock. We may want to do a little averaging across nearby points when calculating the gradient to get the servo gain, because I did see some instances where this seems a little higher than we might want, which could lead to an unstable value being used.
The first graph attached shows two cycles of using guardian to lock the laser, with graphs for the PZT voltage (effectively the control signal) and the power output (effectively the error signal). The second graph shows the data for power as a function of pzt voltage that was collected by guardian during the locking process.
I also ran a step function on the chiller to look at the time constant for the water to reach the laser (see attached graph). It came in at 110s, which is similar to a measurement made at LLO. We will use this to pick the unit gain frequency for the chiller servo loop. The laser took a significantly long time to come to equilibrium, though this is because the step made was large.
Tagging GRD as well
1415 - 1430 hrs. local -> To and from Y-mid Next over fill to be Tuesday, Feb. 2nd before 4:00 pm
Attached plots are the comparison between X04 and C01 data where C01 is latest version of the calibration without the time varying parameters applied to it. X04 is the C01 data with time varying calibration parameters applied to it, except for the cavity pole. These comparsions are made only at the calibration line frequencies and includes data from Sep 30- Oct 9.
X04 agrees closely with PCAL both in phase and magnitude. The magnitude of C01 shows some discrepancy, especially at lower frequency, which can be attributed to the charging of testmass. There is 1 16-kHz cycle discrepancy in phase of C01 as well of which the calibration team is aware of and have plans to correct with the release of X04 (eventual C02).
I removed the mixer-based rfam monitor at ISC-R2 (23567). This was using outputs 7 and 8 of the 45 MHz distribution amp; these outputs are now terminated.
The N connectors on outputs 4, 5, and 6 of the distribution amp were not finger-tight for some reason (perhaps collateral damage from some other 45 MHz investigation?). I tightened them down.
I took step responses for some of our slower ASC loops by applying small offsets to the digital error points. Bandwidths are determined via a simple exponential fit to the error signal response (so in some cases, particularly for the faster loops, this is probably underfitting).
| Pitch [Hz] | Yaw [Hz] | |
| IM4 | 0.10 | 0.06 |
| PRM | 0.03 | 0.2 |
| SRM | 0.14 | 0.5 |
We evidently have great dissimilarity between pitch and yaw for some of these loops.
Times as follows (date is 2015-01-30 Z or 2015-01-31 Z):
I've been working on the Guardian script for locking the CO2 lasers. It now has the following states programmed:
DOWN : Resets the laser pzt position to center. Resets the chiller temperature to 20C. Turns off all servos
FIND_LOCK_POINT : Scans the internal pzt of the laser while measuring the power output using the thermopile. The thermopile is a little slow, so this takes a few seconds to do. It then looks at the slope of the resulting curve to work out if there is a suitable place to lock. It can lock to a positive or negative slope. If there is no good slope then it steps the chiller temperature by 0.03C (we may want to increase this) waits for 4minutes because there is a significant time lag (this may be too short to reach equilibrium temperture) and then retries to find a good point to lock the laser. It finishes by setting a setpoint value for the desired laser power and leaves the PZT at the voltage corresponding to this power.
LOCK_LASER: Engages an integrator between the thermopile readout and the PZT.
RESET_PZT_VOLTAGE : Slowly hands off the DC offset in the PZT voltage from the offset box to the integrator. The offset ends up at 37.5V, which is the centre of the PZT range. This offset plus the integrator output will now correspond to the PZT voltage for the desired laser power.
ENGAGE_CHILLER_SERVO: The voltage from the PZT integrator is used as the error signal to drive the servo for the chiller. The servo is engaged and gains are set.
ISS_ENGAGE : We don't yet need the intensity stabilization system so this state is just bypassed
LASER_UP : This state will be used to monitor the system while in use, looking to see if the power output of the laser exceeds certain boundaries, or whether the laser PZT reaches the limits of its range, which would indicate that the servo has lost lock on the laser.
We still need to setup the gains on the integrator stages for the laser pzt and the chiller. The chiller one is particularly awkward to monitor because the unit gain frequency is so low. The guardian script is currently located in the TCS folder /opt/rtcds/userapps/trunk/tcs/common/guardian/TCS_CO2.py
Tagging GRD
Alastair, Evan
We drove a line at 24.1 Hz in the TCS X AOM to see if it would show up in DARM.
A 0.003 ct excitation into TCS-ITMX_CO2_AOM_OUT_GAIN is enough to create a line in DARM with an SNR of ~5 for a 10-second FFT.
2015-01-30 22:52:15 to 22:57:15 Z.
After resetting all ISI and suspensions to a working state, aligning the IFO and spending two hours to gte DRMI to lock despite a good alignment, I could retuned the P2L of all test masses. I first tried to inject some DHARD_P noise and tune all four gains, but this didn't work out very well.
So I injected noise on the L2_LOCK_P channel of each test mass indipendently, and tuned the P2L gain. The one that got changed most is ETMY (it was compeltely wrong, sign included).
The table below show the old and new values
| Old | New | |
|---|---|---|
| ITMX | 1.848 | 1.850 |
| ITMY | 0.945 | 1.050 |
| ETMX | 1.269 | 1.200 |
| ETMY | 0.433 | -0.090 |
The attached plot shows the transfer function from DHARD_P (measured with noise) before and after. The coupling is reduced by aboput 15 dB above 14 Hz. A bit less below, as usual due to the phase rotation created by the roll mode band stops. In the plot, the two DARM spectra are both with noise injection, same amplitude.
7.0 EQ on Alaskan side of Russia. IFO closed early for the night.
The last day LHO has been suffering from high microseism, so we've had lots of time to ponder ground motion. JeffK noticed a seeming inconsistency on the DetChar summary pages, that maybe someone can clear up for us. The ASD for H1 ground motion last night seems has a microseism peak in Z at about 5 microns, but the BLRMS for that period indicates that the motion is about a micron, a factor of 5 disagreement. Attached pictures are screenshots of the Detchar page for last night, first is the ASD, second is the BLRMS. This could be partly because the two measures are calculated for different measures of amplitude, peak vs peak-to-peak vs RMS. But I think that at best explains a factor of 2 difference, not 5.
The ASD has units of m/rt(Hz), so to get RMS value, you have to multiply by square root of the frequency of interest. So 5 micron/rt(Hz at 0.1 Hz gives an RMS of ~1.5 micron.
There is also the problem of velocity versus displacement measurement, but since f is nearly 0.16 Hz, it doesn't make much difference.
This is just a quick listing of the tests that each IFO has in their DIAG_MAIN Guardians and their docstrings.
LHO
ALS: ALS checks, This can be seen when the beatnote drops down to 0-2 MHz and when (in dBm) it gets above 5. Checks to see if the ALS SHG Temperature Control Servo is on Nominal is 'ON'
BEAM_DIVERTERS: The Beam divereters should be closed after the ISC_LOCK state of CLOSE_BEAM_DIVERTERS #550. 1 == closed, 0 == open.
BECKHOFF: Beckhoff status check
Check some random Beckhoff channels to see if the system in running.
COIL_DRIVERS: Coil drivers functioning, If they stay at 0, it will bring up a notification
ESD_DRIVER: ESD driver status
HWS_SLEDS: HWS SLEDS are on
HW_INJ: Hardware injections running
ISI_RINGUP: ISI ring up motion as seen in alog24245, Only looking at the ETMs and non rotational dofs for now.
MC_WFS: MC WFS are engaged
OMC_PZT: The OMC PZT2 high voltage should be on, otherwise the OMC_LOCK Guardian will fail at the FIND_CARRIER step
OPLEV_SUMS: Optical lever sums nominal, Check only when the suspension is aligned.
PEM_CHANGE: Major environmental changes
PMV_HV: PMC high voltage is nominally around 4V, power glitches and other various issues can shut it off.
PSL_FSS: PSL FSS not oscillating
PSL_ISS: ISS diffracted power
PSL_NOISE_EATER: PSL noise eater engaged
RING_HEATERS: End station ring heaters on
SEI_STATE: SEI systems nominal, Also checks if the WD saturation counts are are greater than somepercent of the limit.
SEI_WD: SEI watchdog not tripped
SERVO_BOARDS: These are Beckhoff controlled channels and cannot be monitored by SDF yet.
SHUTTERS: Monitor the Beckhoff controlled shutters
SUS_PUM_WD: RMS watchdogs
SUS_WD: SUS watchdogs and OSEM input filter inmon on the quads
TCS_LASER: TCS laser OK, Switched ON, Power above 50W, No Flow Alarm, No IR alarm, Only for TCS X for now.
TIDAL_LIMITS: Tidal limits within range
LLO
ALS_LASER: ALS laser power above threshold
ALS_PLL: None
ALS_SERVO_POLARITY: ALS servo polarity monitors
BECKHOFF: BECKHOFF system monitor
BSOPLEV: Beamsplitter OpLev Monitor
IOP_DACKILL_MON: Monitors IOP DACKILL status and reporting faults
ISS_DIFFRACTED_POWER: ISS Diffracted Power monitor
OMC_PZT_HV: OMC PZT HV monitor
PARAMETRIC_INSTABILITY: PI not rung up
PRC_GAIN: PRC gain above threshold
SEI_ISI_T240_SATMON: SEI ISI T240 saturation monitors
SUS_QUAD_ESD_BMON: QUAD BIO ESD driver digital monitors
SUS_QUAD_ESD_DMON: QUAD ESD driver digital monitors
SUS_QUAD_L3_BIO_CHECK: QUAD L3 BIO switch consistency check
SUS_QUAD_PUM_WD: QUAD PUM RMS watchdog
static_vars: None
To get these I opened a python shell and imported inspect and the desired module, then...
for name, obj in inspect.gemembers(module_name):
if inspect.isfunction(obj):
print '{}: {}'.format(name, inspect.getdoc(obj))
Hey, TJ, here's a simpler way, that just prints registered tests and not other functions, for future reference:
servo:~ 0$ guardian -i DIAG_MAIN
--------------------
aLIGO Guardian Shell
--------------------
ezca prefix: H1:
system: DIAG_MAIN (/opt/rtcds/userapps/release/sys/h1/guardian/DIAG_MAIN.py)
In [1]: SYSDIAG.print_tests()
Registered tests:
SEI_STATE: SEI systems nominal
SUS_WD: SUS watchdogs and OSEM input filter inmon on the quads
SEI_WD: SEI watchdog not tripped
HWS_SLEDS: HWS SLEDS are on
PSL_ISS: ISS diffracted power
SHUTTERS: Monitor the Beckhoff controlled shutters
PSL_NOISE_EATER: PSL noise eater engaged
BECKHOFF: Beckhoff status check
SERVO_BOARDS: These are Beckhoff controlled channels and cannot be monitored by SDF yet.
HW_INJ: Hardware injections running
ESD_DRIVER: ESD driver status
RING_HEATERS: End station ring heaters on
ISI_RINGUP: ISI ring up motion as seen in alog24245,
BEAM_DIVERTERS: The Beam divereters should be closed after the ISC_LOCK state of
PEM_CHANGE: Major environmental changes
MC_WFS: MC WFS are engaged
COIL_DRIVERS: Coil drivers functioning
TIDAL_LIMITS: Tidal limits within range
PMV_HV: PMC high voltage is nominally around 4V, power glitches and other
PSL_FSS: PSL FSS not oscillating
TCS_LASER: TCS laser OK
SUS_PUM_WD: RMS watchdogs
OMC_PZT: The OMC PZT2 high voltage should be on, otherwise the OMC_LOCK Guardian will fail
ALS: ALS checks
OPLEV_SUMS: Optical lever sums nominal
In [2]:
The ASC model has been modified, and is ready for the 2f AS "WFS" centering.
The ADC inputs were made to be consistent with T1100472-v16. The spare AS_D (also called AS_X in the model) channels were used, as well as 8 more that had previously been unused. This means that any channels previously called AS_D no longer exist, as they have been replaced with the AS_A_RF90 channels.
So, the model now has 2 WFS demodulator blocks, AS_A_RF90 and AS_B_RF90. The I outputs (pitch and yaw) are sent to the ASC input matrices. Since these are 2f sensors, we can rotate all the signal into the I-phase, so the Q outputs are terminated in the model. These have been put at the end of the matrix so that no channel numbers / names will have changed.
It looks like the DC5 centering loop has always existed, and we just weren't using it, so I didn't add any filter banks for the control loops, nor did the output matrix change in the model. Daniel has already added the DC5 control filters and the DC5 matrix column to the medm screens.
The model compiles, and has been checked in to the svn. While I still have other ASC model modifications, this version should be ready for install during Tuesday maintenence if I don't get the other modifications done in time.
Medm screens have also been modified. They have white boxes for now that will be populated when we restart the model on Tues. I think we're ready to roll software-wise for the 2f centering.
