This evening Lisa pointed out that the low carrier buildup can be explained by misalignment in the Y arm.
I spent an hour to carefully adjust the alignment. Indeed, after the alignment, I got a cavity flash of 45 (normalized for a single arm build up) which is the highest we ever obtained. Also, the funny REFL9I offset disappeared and I could see a PDH-looking signal there when CARM passed through a resonance.
What I did:
This is really the next big watchdog issue: regular IFO commissioning activity is causing the SEI watchdogs to trip. We need to investigate the issue further, but what we definitely know is that when we lose the ALS locks the ETM ISI watchdogs trip. It's usually just in stage 2 actuators, but also sometimes both ISI stages and more than just the actuators. This is reducing efficiency during commissioning, so we need to back off on the saturation count for the SEI watchdogs some more.
Do you know why the SUS watchdogs don't trip before the ISI watchdogs? I would imagine that the transient sent when we lose the lock would trip SUS first.
SR3 measurements will be running overnight on opsws1.
Guardian states "undamping/undamped" were added to sus.py. It basically set the correct switches for testing purposes (undamped tfs). This state can be called from matlab before running a TF using (e.g for SR3):
system('caput H1:GRD-SUS_SR3_REQUEST UNDAMPED')
Only SR3 guardian was restarted
Measurements still running this morning
Greg, Thomas, David Corner HWS table: - both end-X and end-Y in-air optics have been aligned (using the fibre coupled 532nm source) from the source to the periscope. Slight tweaking may be required once the SLED is connected up. - air lens and collimator temperature sensors for both end-X and end-Y are connected up and are working - HWS for both end-X and end-Y are yet to be assembled/installed - image of the corner HWS table attached TCSX: - laser has been operating all day. Thomas working on connecting up FLIR camera computer. - laser will be left running again overnight to allow performance monitoring TCSY: - alignment confirmed through the AOM, while the picomotor mirrors were repositioned. Alignment is now through the second pick off optic, which is just before the HWP and polarisers. - CP power control polarisers were removed and checked for pitching using the reflection from a HeNe (set up to be at a flat 4" beam height). Initially the first polariser had pitched the beam down 0.25" at a distance of 20". This pitching was removed by loosening and retightening the bolts that secure the polariser mount to the mounting bracket. Polariser plate was also observed to move in this mount, further changing this pitch. This polariser plate is secured in the mount using a nylon screw, and when tightened improves this a bit. Second polariser appeared to be slighly better. Further investigation required.
A representative from Electric Utilities Dispatch called the control room to report that they are having power issues, but he stressed that we should not be affected. He did mention, though, that if we experienced and power glitches today they could have been associated with the fault that they experienced.
He signed off by saying "I hope you understand the gravity of the situation", followed by a brief pause, then harty laughter. Thos jokers...
F. Clara, R. McCarthy and R. Weiss We used the capacitance bridge circuit to establish the wiring of the ETMY. We found (as Keita had guessed) that the common bias line was connected to cable 1 and not cable 3. The capacitance change between the bias trace and the individual quadrant control traces was measured as the recoil mass and the test mass pendulated driven by the ground noise. The seismic system and all suspension damping was turned off during the measurement. The accompanying figure shows the voltage spectrum at the output of the lock-in amplifier for each quadrant. The number at the right is the amplitude of the pendulum peak for each quadrant. The smaller value for the capacitance change between cable 1 and 5 is somewhat worrisome. If this measurement needs to be repeated to validate a difference in the force/voltage^2 ratio for the quadrants, it would be best to deliberately apply a force to the penultimate mass at the resonance frequency of the pendulation rather than to rely on the ground noise. The peak measured in all the quadrants is at 0.4297 Hz
Yesterday I edited the ALS diff guardian in order to add new states which automatically find an IR resonance. The requestable state is called IR_FOUND. The concept is the same as that of ALS comm. It tries two specific DARM OFFSET places and checks the arm build up in Y arm. If it doesn't find a high buildup, it times out and falls to a failure state, called NO_IR_FOUND. At this point the user should manually find a resonance to get out of this state.
The power code for the EY laser was found to be disconnected. According to a trend, this happened at 9:20 AM this morning. Probably people working around there accidentally disconnected it.
I made a new BSC-ISI overview screen, with a related display, to support Fabrice's new Z drive to RZ T240 decoupling.
The use of the feature requires a model change which has only been installed on BSC-ITMY so far. It will be installed on the other units uppon re-compile/restart once the related ECR is filed and approved. Hence, the sitemap was modified to load the new display for ITMY only -r8044
The new BSC-ISI MEDM screens are commited in the svn:
opt/rtcs/userapps/trunk/isi/common/medm -r8045
For some reason, the fonts look good on the editor, but not in run-mode (see attached screenshot)
8:38 am, David H. to CS VEA, Hartman table work.
8:53 am, Betsy to CS VEA, looking for parts.
9:00 am, Richard, Rai and Filiberto to Y-End VEA, ESD testing.
9:30 am, Keita and Daniel report a moth problem at X-End Station, no success on trapping the insects.
10:05 am, Cyrus and Adrian to Y-End VEA, upgrade CDS computers.
10:35 am, Bubba and crew CS VEA, remove HAM arm from HAM5 north entrance.
10:45 am, Betsy to CS VEA, West bay area looking for parts.
11:45 am, Cyrus and Adrian to Y-End VEA, install mice on CDS machines.
1:29 pm, Richard, Rai and Filiberto to Y-End VEA, ESD work.
2:47 pm, Received a call from power company, power glitch/outage on grid, reported to Richard M.
3:10 pm, Filiberto moving ESD test equipment from Y-End station to X-End station.
3:30 pm, Cyrus and Adrian to H1CER, upgrade CDS computer.
4:15 pm, Kiwamu to Y-End VEA, checking laser status, no light.
Did the same thing as X (3Hz injection to ESD each quadrant, with various offsets in DC and quadrants, measured Oplev response). And it didn't make sense (first attachment), nothing looked like a nice simple function of voltage difference.
That's until you notice that at least UL, UR and LR behave as if the bias voltage stayed the same for the entire measurement. In the second plot, I used the same data as the first one but assumed that the bias stayed at zero volt, and you see that UL, UR and LR looked quite nice. The only odd thing is LL.
So I assumed that LL was somehow cross-wired to the DC bias. I give LL -377, 0 or 377 volt, and repeated the same measurement (I used "DC" as a replacement for LL).
This time it makes much more sense (3rd plot). Comparing it with X ESD, assuming that the oplev sign convention is the same for X and Y (which should be the case), the conclusion is that:
Or, if the oplev sign for P is the same but Y is the opposite of EX (which should not be the case)
Anyway, the effective bias voltage etc. are:
UL | DC | UR | LR | |
P charge effective voltage (V) | 138 | -842 | -53 | 58 |
P slope (urad/V^2) | -7.7e-8 | 4.7e-8 | 6.2e-8 | -7.9e-8 |
Y charge effective voltage (V) | 73 | -460 | -52 | 45 |
Y slope (urad/V^2) | -5.7e-8 | -5.2e-8 | 6.1e-8 | 6.3e-8 |
Seems like the slopes are about the same as EX though DC is somewhat weaker.
Also, DC has a crazy offset which, if due to the charging, is equivalent of -842V (PIT) and -460V (YAW).
I manually measured the equivalent bias voltage of ESD caused by the charging for ETMX by looking at the oplev.
I changed the bias voltage (-377V, 0, 377V) and measured the transfer coefficients from the drive to the quadrant to the oplev at 3Hz.
In the case of UL and LL, I also put some offset in the quadrant output so I have a larger dc voltage difference between the bias and the quadrant.
As you can see, though it's not a perfect line as you can see in UL and LL, it looks very reasonable in that the absolute value of the slope of four quadrants are quite similar, plus PIT/YAW makes perfect sense.
The effective bias voltage obtained from PIT data and YAW data are different because it is dependent on where the charge is.
For each quadrant I fit the three data points that correspond to no offset in the quadrant itself and obtained these:
UL | LL | UR | LR | |
P charge effective bias [V] | -161 | 78 | -176 | 43 |
P slope [urad/V^2] | 7.6e-8 | -7.0e-8 | 7.8e-8 | -6.0e-8 |
Y charge effective bias [V] | -201 | -60 | -233 | -88 |
Y slope [urad/V^2] | 7.3e-8 | 7.3e-8 | -6.5e-8 | -6.9e-8 |
I found ETMY-ISI drive going up to huge numbers with the masterswitch off. Nothing was going out to the DAC since the masterswitch was off, but the Guardian should have turned the ISO loops off to prevent sending a big kick into the ISI when the master switch is turned back on. Jamie is going to provide a fix for this: The ISI will go into the "Offline" state, and turn back on from there, anytime its masterswitch is truned off.
As a side note, we should not turn the ISI off using the masterswitch while it is running. In this situation, the "Offline" state should have been requested from the ISI's Guardian menu.
I tracked down this issue to a couple of problems/inadequecies with how the SEI guardian was handling the masterswitch.
First of all, the "masterswitch_is_on" decorator, which is used in all states to check the masterswitch status, had been disabled. This was probably done during debugging and was forgotten. I re-enabled the check so that the ISIs and HPIs now jump to their MASTERSWITCH_OFF states when the masterswitch is off.
I then fixed the MASTERSWITCH_OFF states so that they immediately shutoff all isolation and damping loops.
I finally added masterswitch status checking in the SEI managers, so that they handle errant masterswitch state changes more cleanly.
This all means that the the guardians should have much cleaner handling of the masterswitch being turned off.
That said, and as Hugo mentioned, don't turn off the masterswitch. This is a rather violent thing to do to the ISIs, as they'll get a large jolt if any DC offsets are suddenly removed. Instead, use the SEI managers to request the OFFLINE state.
Daniel and I restarted green WFS effort.
WFSA and WFSB were rephased using an injection into PDH CM board. Apparently somebody tried to compensate the signal differences across the quadrants when doing a similar injection as mine, and I followed that same path such that seg1 and seg3 are matched as well as seg2 and seg4.
Then I dithered PZTs at 2, 8, 11 and 13.5Hz to measure the sensing matrix, inverted the matrix, and put them in the input matrix, such that WFS_DOF3 becomes PZT1 and DOF4 PZT2. Note that both WFSs are much more sensitive to PZT2 than PZT1.
I didn't put any filter in DOF3/4 nor IP PITOFS and IP YAWOFS, but set the gain to some negative numbers.
In the input matrix of QPD alignment servo, I added a straight-through matrix to add the WFS signal.
I had a hard time working with state definition thing and maybe the guardian, so I still don't know if it works or not, but the intent is to do a high BW feedback to the PZT.
This was due to the guardian still containing old WFS code. I removed the offending lines from the python script. Then, I reloaded the guardian. Unfortunately, I missed adding a pass statement in a hanging if clause, which caused the guardian to commit suicide. Unfortunately, the guardian took the restart button with it. No guardian, no restart button, ???
Cyrus, Adrian
We've changed the OS on the workstations at EY (eyws0, wyws1) from Mac OS to Linux, matching the current setup at EX (now that the bugs seem to be mostly worked out).
Also did the same to lveaws0 in the CER.
(John W., Gerardo M.)
The cold cathode was turned off in an attempt to extend its life.
Pressure before turning it off was 6.10x10^-05 torr.
The CC will be turned on and reading will be posted to keep track of the pressure.
However the CC will be off over the weekend.
Turned the CC at 3:00 pm local time, waited for 15 minutes then took a reading, 6.05x10-05 torr, I turned the CC off.
Minutes taken by Corey G.
alog down
TCS: have light on cameras, hartman work this morning & laser hazard in afternoon
HAM4
OFI: installed & roughly located
Apollo: request to remove HAM4 installation arm when convenient
Baffle installation remains
SEI: balanced, may be unlocking HEPI soon
HAM5
sei awaiting final payload
HAM6
sled work in Optics Lab
will be Laser Hazard intermittently
ACB finished yesterday, will start CPB
Apollo: HEPI plumbing
EY ESD testing (swinging etmy) this morning Rai/Richard/Filiberto
ISS PD Array cont. (Peter King)
Also in the afternonn I did a simple, brief chek for the demod phase of all the REFL 1f PDs.
I swept the X arm through a resonance by using ALS comm and monitored the PDH signals. It seems that everyone is adjusted somewhat.
=== settings ===
REFL9 = 84.9 deg
REFL45 = 19.2 deg
REFLAIR9 = 90 deg
REFLAIR45 = 147.6 deg
The whitening gain for all the PDs was set to the maximum i.e. 45 dB in this measurement.
This is a summary of last week of locking work, for posterity.
https://dcc.ligo.org/LIGO-G1400576