On friday night, the wind was high(~25mph), and the arm was hard to lock. Although the angular motion of the test masses was not particularly different than on a quiet night. First attachement shows the oplev spectra from friday night, when the wind was ~25mph. second attachement shows spectra from yesterday night with a 5mph wind.

To note :

• RMS of ETMY (blue curve) in pitch and yaw are consistently higher, even though the ISIs are in the same config (level 3 controller and Tbetter blends on St1 and St2 except for Rx and Rz)
• ITMX (pink curve) seems to perform better, but it's actually because the oplev beam is offcentered from the qpd, sitting mostly on S4 quadrant. SUM is ~5900

(Stefan, Lisa, Chris)

Looking through the locked stretches from the weekend, we noticed that the ALS slow feedback -- intended to relieve the VCO control signal of low frequency seismic and tidal motion -- was failing in its mission.  What we saw is plotted in the first attachment.  The green trace shows the VCO control signal, which should be forced toward zero when the slow test-mass feedback (red trace) is enabled about halfway through the timeseries. Obviously this wasn't happening. Meanwhile the VCO frequency servo that actuates the tune slider (blue trace) was running away along with the VCO control signal. When the tune slider hits its limit at +/-5 the lock is broken.

We tracked it down to a missing integrator in the ALS-X_ARM filter bank, without which the slow loop didn't have enough oomph at DC to null the fast control signal. I think this filter was removed intentionally (but without realizing its true purpose) during some earlier automation work. We put it back and, taking advantage of the improved test mass plant inversion, we were able to increase the gain upstream (in the ALS-X_REFL_SLOW filter bank) from 1 to 30. This should help prevent the arm from being blown away by the wind.  When we lose lock the integrator is switched off by the Beckhoff, and the drive bleeds away through a 0.01Hz pole.  The improved behavior is shown in the second attachment (where the gain was cranked about 20 minutes in).

Here is the list of commissioning task for the next 7-14 days:

Green team (XY-arm):

• Hand-off to 3f lock signals.
• Common mode hand-off.
• Hand-off ALS signals to interferometer signals.
• RF phase for MICH.
• Add CM Sum in chain.
• Noise hunting.

Blue team (X/Y-arm):

• Revisit the WFS with the now improved beam quality.
• High BW feedback to input PZTs?
• EY WFS alignment sening matrix.

Red team:

• Center the red QPDs in EX/EY.

SEI/SUS team:

• EY ESD drive is broken.

In-Vacuum preperation:

• IMC/PRCL adjustment: Prepare to move MC2.
• EY ESD fix: What measurements can we do now?
• Conflat for mirror discharging.
• EX QPD centering: Can we get the beam centered?
• EY QPD centering: Will work.
• POP beamsplitter.
• Change mode of ALS/POP periscope in HAM1.
• Are all picos working?
• Test photodetector pods for leaking?

As of 12:00

PSL Status: 
SysStat: Green, except VB
Output power: 26.7w  
FRONTEND WATCH: Good
HPO WATCH: Red  

PMC:
Locked: 12min
Reflected power:   1.5w
Power Transmitted: 9.8w 
Total Power:      11.3w 

FSS:
Locked: 12min
Trans PD: 0.94v

ISS:
Diffracted power: 3.833%
Last saturation event: 12m

Jason & Betsy – SRM fine alignment at HAM5
Gerardo – Putting the OFI into storage box and fly over beam tube
Hugh – ISI balancing at HAM4
Travis – ACB balance and alignment
Richard & Filiberto – Chassis work in EE shop & cabling at HAM6
Peter & Olli – In H2-PSL working on ISS assembly
Dave – TCS Hartman table alignment

The problems we noted over the weekend with the ETM SUS (and sometimes SEI) systems tripping on lock loss (alog 11957) can maybe most accurately be characterized as a watchdog problem.  The trips are caused by large drive transients after lock loss.  It's hard to imagine that there's much we can do to prevent this from happening.  Even if the DARM (CARM, etc.) control signals could be shut off immediately, the residual impulses would still produce large impulse response in the LOCK filters in the suspensions.    We could try to shut off the drive signals in the SUS controllers, or hold the outputs at their current value, but that's a bit more difficult to implement.

In general, though, the watchdogs should probably just not be tripping on transients.  If the watchdogs were a bit smarter and only tripped on sustained saturations or oscillations, this would likely not be an issue.  I vote that we solve the problem in the watchdogs, but increasing the amount of time it takes before they trip.

Just for a bookkeeping purpose:

I recentered the beam on the IMC WFSs on this past Saturday. I used the picomotors. Both of them had been off by 0.2-ish counts in the normalized pitch and yaw.

No restarts reported.

I made a couple of improvements to the IMC guardian:

• maybe most importantly, I filled in a lot of comments and inline documentation
• cleaned up the code quite a bit
• added GuardStateDecorators for handling fault conditions and lock loss
• cleaned up edges around the FAULT state
• broke out a BOOST state from the LOCKED state

The final point was to make it so that there is no activity in the LOCKED state, so that reaching the LOCKED state means that the IMC is fully up.  In general, I think we should start making all requestable states be "idle" states in that they don't do any action other than monitoring for exit conditions.  This idea here is to make them true markers of a steady state of the system.

We end up with just two requestable states now: DOWN and LOCKED.  The DOWN state doesn't actually prevent the IMC from locking briefly, so we may want to change the DOWN state to something that actually prevents the IMC from locking.

Next, I want to make the IMC guardian the manager of the SUS_MC{1,2,3}, so IMC is actually in control of setting the MC suspensions to their ALIGNED states.  This will also allow us to achive the point above, since we can then make the DOWN state into something where the SUS are misaligned, thereby preventing the IMC from locking.

model restarts logged for Fri 16/May/2014
2014_05_16 23:12 h1fw1

No restarts reported for Saturday.

One unexpected restart of h1fw1 over the two days.

Kiwamu, Jamie, Lisa

We figured out why we couldn't reduce the CARM offset below 1kHz while keeping the PRMI locked on REFL 45 I&Q.

The REFL 45 demod phase was tuned without the arms present; it turns out that with the arms, even with a large CARM offset (1 kHz), the demod phase changes enough to make MICH very close to instability.

So, we just tuned the REFL 45 phase by monitored the MICH OL TF and the REFL 45 I/Q decoupling by looking at a PRCL line around 50 Hz.

Not a very advanced technique, but we could recover pretty easily a reasonable shape for the MICH loop and we could reduce the CARM offset.

Kiwamu will post some plots with the MICH OL TF; here are some numbers for the REFL 45 phase:

Without the arms: 144 dg

1 kHz --- 900 Hz CARM offset: 172 dg

800 Hz  --- 700 Hz CARM offset: 180 dg

We were happy and ready to do the transition to 3f, but another seismic event like the previous one made the lock of the Y arm more difficult. 

We are still debating if we should go home or not...

[Jamie, Kiwamu, Lisa]

The ETM SUS are getting kicked to the point of watchdog trips when the we lose the ALS DARM lock.  The DOWN state of the ALS_DIFF guardian was shutting off the INPUT to the LSC-DARM filter bank, but not the OUTPUT, so we were suspecting that leftover filter stuff was being driven out to the suspensions.  There are also no LSC triggers associated with the ALS error signals that could shut off the DARM output after lock loss.

I modified the ALS_DIFF DOWN state to turn off the DARM output as well, in the hopes that that would reduce the problem.  We thought that that did help, but then we just got another lock loss that did trip the ETMY SUS and ISI ST2 watchdogs, so it didn't elliminate the problem entirely.

I wonder if we'll need to add some LSC triggering based on the ALS inputs, for when we're using ALS signals to feed back to the SUS.

While I was at it, I did a bunch of code cleanup in the ALS_DIFF module.  Mostly just to keep things consistent with the guardian style guide (unpublished).  I also committed the ALS_DIFF.py module to the USERAPPS svn, and added it to the GUARD_OVERVIEW medm screen.
 

Kiwamu, Jamie, Lisa

Today we had the illusion that we could start locking before dinner time, but that didn't happen.

While both the X and Y arms were locked on green, something happened in the corner station (visible in the PEM CS SEIS 30-100 mHz channels, especially Z) which affected only the Y arm, while the X arm remained happily locked.

Lisa, ChrisW, Jamie, Arnaud, Kiwamu

We locked the PRMI with the 1f signals at a CARM offset of 1-2 kHz. The goal is to hand it over to the 3f after the CARM offset passes approximately 1 kHz at which the arm cavities resonate with the 2f sidebands. However, we didn't get a real progress tonight. We will continue studying this configuration.

What we did:

• We aligned TMSX by using the usual baffle diodes technique.
• This resulted in a small beatnote at the ALS diff broadband PD. So I realigned the beatnote.
• The SHG was found to be somewhat low. So I realigned the crystal. Now the power at 532 nm is about 0.024 mW according to the monitor diode. This is a factor of two improvements. Good. The comm beatnote is now at -1 dBm at the PFD monitor.
• We had a difficulty in engaging ALS diff.
  • This turned out to be due to new L2P filters in UIM of ETMY that Aranud put in today. Although the filter is more accurate, the impulse response in pitch was the killer when the DIFF loop is being engaged. So we went back to the legacy filters. This fixed the problem.
  • Also, the DC bias of ETMX ESD had an opposite sign. This prevented us from going to a high UGF. So we flipped it back to 125 kcnts as usual.
• There was a high-cavity-motion time in which we were not able to hold the green PDH for 1 min due to the PDH actuation range.
• We started using the slow feedback onto ETMs.
  • Since the guardian codes tend to clear the length history, this produced large kick on ETMs every time when it looses the lock.
  • We commented out the sentences which clear the history in order to reduce the kick.
  • Also, I changed ALS-X(Y)_ARM filters because the PDH autolocker was clearing the history by toggling the first two filters. Right now these filters are intentionally left empty such that the autolocker does not clear the history.
• At some point, we noticed that the cavity stay locked without the slow feedback. So we stopped using the slow feedback.
• After some struggling to align the PRMI, we managed to start the real experiments.
• I tried CARM offsets of 1 kHz and 2 kHz. These worked just fine.
  • While holing the PRMI with the 1f signals, I tried to approach the resonance by reducing the CARM offset to see how it behaves.
  • In the case of the 2 kHz offset, it dropped the lock when the offset was about 1.5 kHz.
  • In the case of the 1 kHz offset, it dropped the lock when the offset was about 900 Hz. Both lock losses seem to be associated with high frequency contamination. I didn't take a close look yet.

The coupling from the BSC-ISI Stage 1 Z drive to the T240 RZ signal reported a few weeks ago has been lately our first guess as for what's limiting the RZ (Yaw) isolation performance.

This week we modified our real time model to do subtraction tests. The modification has been installed on ITMY.

The first plot attached shows:

- the coupling measurement in green

- a simple fit to be used in the subtraction block, in black

- the expected residual coupling in red

The second plot shows the residual coupling measured after installing the correction. The subtraction results look pretty good. Unfortunately, I did not observe substantial improvement in the optical lever Yaw motion. To be further investigated...

Came in this morning to find the BS ISI stage 2 GS13 watchdog tripped.  Unclear why.

The strange thing is that the trip appears to have occured at GPS 1084253752, which is 10:30pm last night, whereas commissioning finished up around 4:30am.  Did they just run with the BS tripped all night?

The watchdog was reset and guardian recovered everything fine.