Late last week we had one of the two compressors quit on the AAON unit at the staging building. See FRS 3455. This unit is designed to supplement the original 12 T Carrier chiller which also has one of the two condenser fan motors beginning to show signs of failure. See FRS 3451.

The failed compressor has been disconnected and the AAON is operating at 50% capacity which should be enough to assist the main unit enough to keep the staging building cool especially during the somewhat cooler temperatures.

The faulty parts have now been ordered and hopefully arrive before any additional triple digit temperatures.

 

LVEA: Laser Hazard
IFO: UnLocked
Observation Bit: Commissioning  

All Times in UTC
07:00 Take over from Ed M.
07:00 Commissioners working on IFO
09:11 Locked at Nominal_Low_Noise – Commissioning on going 
11:09 Lock Loss – Commissioning activities
11:30 Dust counts at End-Y high
12:01 Locked at Nominal_Low_Noise – Commissioning on going 
15:00 Turn over to Travis
15:04 Lockloss - 

The attached pdf contains current Stage-1 and Stage-2 NB model performance of ITMY chamber for X,Y,Z and RZ dof. 

Fig 1- ST1 ITMY X
Low frequency near microseism model performance is limited by the ground blend filter (HEPI L4C+IPS)
1-5Hz frequency band is limited by GND-STS passing through the sensor correction path
5Hz and above (upto 10Hz say) is limited by Stage-2 back reaction
Below microseism model does not match with actual performance.       (May be because of some tilt coupling??  But it looks like actual measurement is limited by the sensor noises (T240/CPS) via isolation filter. - Not sure about it)

Fig 2- ST2 ITMY X
Instead of T240 BLND OUT,  I have used T240 BLND IN as the input (stage-1 displacement) to the ST-2 model.
Though the model and measured GS13 are in agreement above blend frequency (250mHz), the shape between 1-3Hz are not same. The model output looks more like the input signal T240 BLND IN.  May be a better input noise model will work better. 
Since the stage-2 model performance above ~1Hz is highly dependent on stage-1 displacement, we can san say that the Stage-2 back reaction on stage-1 can have effect on the final performance of the model. 
Fig 3 - ST1 ITMY Y

Most of the features and model performance are same as X dof.

Fig 4 - ST2 ITMY Y

Though this one is same as ST2 ITMY X only thing I have noticed here is the actual IFO ST1 performance is better than ST2  between ~ 300-500 mHz. 

Fig 5- ST1 ITMY Z

ST-1 model performance matches the actual measurement at almost all the frequency range of interest.  
The conclusions derived for ground model and Stage-2 back reaction hold here too.   

Fig 6- ST2 ITMY Z

Unlike X and Y dof, the model performance between 60 to 250 mHz is quite good (I am still trying to figure out why this sort of discrepancy exists between these dofs ???)
At the same time the mismatch between model and actual performance above 5Hz is noticeable. Model is over estimating the actual performance here (though the model has already included stage-2 back reaction in Stage-1)
Fig 7- ST1 ITMY RZ

Apart from the limitations of the model due to ground noise at low frequencies, the performance of this stage is mostly limited by CPS sensor noise via BLEND+ISO  path.

Fig 8- ST1 ITMY RZ
   
Please DO NOT go through this figure. Still need to sort out the problems.

   
Same sort of features can be seen in almost all the BSC chambers except BS where the stage-2 controllers are not in use. 

Have been seeing high dust counts at End-Y during evening ops shift. The counts are not that far off the normal values for End-Y, but there is no apparent cause for the spikes. Will continue to monitor.  

Darkhan, Jenne, Stefan, Evan

Upon revisiting the shape of the DARM loop, we found that we had 4 dB of gain peaking from 10 Hz to 50 Hz.

That seems a bit too strong for our taste, so we took a look at the DARM suscomp filter that we've been using for the past few months. Aside from the low-frequency (<2 Hz) suspension resonance compensation and the high-frequency (>900 Hz) inversion rolloff, there is a single pole at 200 Hz.

This serves us well when locking DIFF, but in full, low-noise lock it costs us phase (not least because we also have the effect of the 350 Hz RSE pole).

Anyway, there is now a filter installed in EY L3 lock with a zero at 200 Hz and a pole at 1000 Hz in order to push this pole up to higher frequency. This gives us a modest improvement in phase margin at the DARM ugf (34° to 46°). The DARM ugf is 40 Hz or so (more or less the same as before). The EY ESD has about 16000 ct rms, mostly accumulated around 1 kHz (magenta/orange in the attached spectra show the before/after).

[In the longer term, we should just roll p/z pair into the suscomp filter, so that it rises like f essentially from 1 Hz to 1 kHz. Then put a 200 Hz pole / 1000 Hz zero pair in EX L3 lock, so as not to disturb the DIFF loop shape. However, I've put what we have so far into the guardian and it works fine.]

Darkhan and I retuned the DARM OLTF template in order to not saturate with this new loop shape. The attachment shows the before (blue) and after (red), along with the closed-loop gain and loop suppression. There's still moderate gain peaking below 20 Hz.

Evan, Stefan,

Evan will make a more detailed log entry with actual measurements, but here are the highlights:
- We beat the 45Mhz signal from the installed and the spare harmonic generators.
- the first odd thing was that straight out of the harmonic generators, the 45Mhz signals were out of phase by about 170deg, even though the 9MHz inputs were nicely in phase...
- We phased the two signals to be exactly in phase an stuck them into a mixer: 0.415V
- We phased the. To be exactly 90deg apart: no DC signal, and ~ 220nV/rtHz of flat  broadband noise.
- this gives 5.3e-7 rad/rtHz between the two signals.
- Assuming equal and non-coherent contributions, the phase noise of one box is thus: 3.75e-7 rad/rtHz
- If we assume that the oscillator phase noise has the same coupling as the oscillator amplitude noise (9e-2mA/RIN), we would expect 3.4e-8mA/rtHz. Thats's pretty close to what we see on ODC_DCPD_SUM.

We had noticed last week that the refl 45 WFS were phased incorrectly, and tonight Jenne and I rephased them by moving PR3 100 counts in pitch.  This only affects the PRC2 loop.  We've locked and engaged the ASC several times since then, and this seems fine.  

While we did improve the phasing and tuned it to within 5 degrees, I is only about 9 dB higher than Q for PR3 motion.  We will need a different phasing strategy, where we phase 45 to minimize the signal from chard in one quadrature if we want to distinguish between PR3 motion and CHARD.

The first two attached screenshots are the phases before we started, except that for REFL A the phase of the first quadrant was -165 not -145.  The following 2 screenshots were the phases we rephased.  For the time being we have reverted. 

Jenne and I spent some time making SDF green before the enigneering run starts in the morning.  

The main things that are still showing diffs are HAM1 HPI, some violin mode damping settings on ITMY, and many diffs in CALCS.  

While we were doing this we noticed a potentially nasty trick that SDF does.  We chose some channels that should be not monitored from the list of DIFFs, and right as I was about to CONFIRM, the IFO lost lock.  A new diff showed up on the top of the list, and the channels I had intended to not mon were moved down the list, but the yellow squares stayed in the same place.  It would have been verry easy to miss this and apply the action to the wrong channels.  Jenne is putting this in bugzilla.  

The OMC_LOCK guardian now has the states ADD_WHITENING and REMOVE_WHITENING, which do what they say on the tin. If you have one stage of whitening on, running ADD_WHITENING will turn on one more stage, etc.

They are not connected to the rest of the graph at all, so you'll have to run them manually.

They are intended to be compatible with full lock, so they can be run at any time (so long as you don't end up saturating the DCPD ADCs, of course).

ALL TIMES IN UTC

Arrived to find IFO beginning lock sequence after TJ did some PRMI alignment to help DRMI. It seems previous lock segments were on the average of about half a hour

• 23:09 DRMI locked.

• 23:10 Robert in the LVEA

• 23:15 Guardian hung on an OMC error. (no light on QPD)

• 23:37 noticed HEPI L4 WD SATURATION MONITOR for ITMX is at 43820cts. WD limit is set to 30000(safe)(max is 122880). WIll reset at next opportunity.

• 23:29 Dick and Wayne in the optics lab

• 23:30 Robert out of the LVEA

• 23:48 reset ITMX HEPI WD accumulator.

• 00:10 DRMI taking in excess of 16 minutes. No environmental noise observed. Tried PRMI lost lock. restored slider values back 1 hour. Re-Aligned SRM DRMI locked 1 minute later.

• 00:29 Wayne out of optics lab.

• 01:02 Stefan out to CER

• 01:16 Dick out

• -1:20 Guardian ISC_LOCK in Error at Engage ASC Part 2. Stefan cleared error

LOCK LOG:

• 23:23 Locked at NLN

  • 23:47 Lockloss at ENGAGE_ISS_SECOND_LOOP

    • DRMI ~ 13 minutes to lock

• 23:49 Began sequence

  • DRMI taking over 16 minutes.

  • 00:20 PRMI align

• 00:48 Locked at NOMINAL_LOW_NOISE

  • 0:48 Lockloss . Gabriele trying to fine tune OMC

• 00:49 Began Sequence

  • 01:07 Lockloss @ ENGAGE_ASC_PART3

• 01:08 Began Sequence

• 01:29 Locked at NOMINAL_LOW_NOISE

• 00:00 IFO still locked. 70mpc

Evening Summary:

• IFO locking in half hour segments. DRMI was taking a rather long time locking. Visiting PRMI may have improved it slightly.

• Stefan and Evan have been taking measurements and tuning for the last 5.5 hour lock.

• IFO still locked at shift change @ ~70-72mpc.

The inspiral range had trended down significantly in the last ~2h of the lock.

- Looking back at the data the spectrum showed a 1/f^2 noise floor between 15 and 150Hz. (plot 1)
- This floor was clearly generated by frequent glitches. Plot 2 shows a filtered time series for the bad time (blue, 2015/08/16 15:05 UTC) and a good time (red, 2015/08/16 13:35 UTC)
  Filter: zpk([0;0;],[1000;1000],1,"n")ellip("BandPass",6,1,80,30,200)

- Sheila noticed that both the M2 and M3 coils of the SRM were hovering around the ominous 65536cts mark, so there is the suspicion that these are SRM glitches. However I couldn't line them up convincingly... Either way, Sheila did the top stage bleed-off today, so we will see...

Evan, Sheila, Jenne, Stefan

we have been intending to offload the SRM length control to M1 (instead of M2) for some time now.  This is now done in the guardian, and is basically a copy of what was done in alog 19850.  The only difference is a 50Hz Cheby low pass, which is used as a cut off instead of a 70 Hz elliptic.  This is in the ISC_DRMI guardian now. 

I have turned off the L2P and L2Y decoupling, as I could only use very low gain with them on.  

Update:

Neither the PRM nor SRM had a true integrator in the top stage, now they both have one which we have engaged by hand, and the guardian has been edited to turn these on, which we will test after our next lockloss. 

Last update: We've tested the integrator switching on now, its fine.

Sheila, Evan

Running the dc-coupled oplev servo on SR3 for long periods of time seems to be causing us grief (LHO#20519).

As an alternative, we are trying out a cage servo which feeds the witness sensor in pitch back to M2:

ezcaservo -r H1:SUS-SR3_M3_WIT_PMON -g -300 -s <SR3 M3 P witness setpoint> H1:SUS-SR3_M2_TEST_P_OFFSET

The step response of this loop has a time constant of 30 s or so.

I made the suggestion before, but today I actually tried it:

By driving both L2 and L3 at a fixed frequency, but with a gain and phase such that their contribution cancelers in DARM, we can easily monitor (and servo, if we want) the ESD drive strength, which is expected to vary with charging.

This method has the big advantage that there are NO LARGE cal line resulting in DARM.

Here are the settings I tried:
- H1:SUS-ETMY_L2_TEST_L_EXC: drive with 300 cts at 20Hz, phase   0  deg
- H1:SUS-ETMY_L3_TEST_L_EXC: drive with 51.7cts at 20Hz, phase 134.2deg

The resulting line in DARM is only ~0.016 times the size of the line with only 1 drive. I also used the H1:SUS-ETMY_L2_DAMP_MODE7_BL filter bank to monitor the line strength. This suggests we should get on the order of 1% monitoring precision on a few second time scale - all while only producing the tiniest peak in DARM - with no up-conversion feet.

Also - since the biggest variation we see is on the microseism time scale - we should try the ESD linearization..

I alrady noticed yesterday that the DARM noise at the periscope peaks (200-400 Hz) was high at the beginning of the lock and then reduced over time.

The first attached plot shows a BLRMS of DARM around those peaks, starting right after reaching the low noise state. There is a clear reduction of the noise over time. The second plot shows that on a similar time scale, the OMC alignment output signal changed, mostly ANG_Y.

This seems to confirm the idea that input beam jitter at the periscope peaks is converted into intensity noise by an OMC misalignment, which changes over time.

To confirm this, I move the OMC angular loops during full lock, adding offsets of few tens of counts to the POS and ANG loops. The third plot shows the steps in the control signals. I was able to reduce the BLRMS by adding an offset of -40 to the ANG_Y loop. The fourth plot compares the DARM spectrum with (red) and without (blue) the ANG_Y offset. I should have increased the offset more. Unfortunately, I noticed that the output was hitting the limit of 300 and I stupidly increased it to 1000: but since the loop was integrating, I broke the lock since I suddenly increased the output from -300 to -1000.

However, the experiment confirms that the noise at the periscope peaks changes in amplitude with the OMC alignment, which is not optimal.