Laser Status:
Front End Power is 33.4W (should be around 30 W)
70W Output Power is 75.16W
Front End Watch is GREEN
70W Watch is GREEN

PMC:
It has been locked 1 days, 19 hr 33 minutes (should be days/weeks)
Reflected power = 17.05Watts
Transmitted power = 43.55Watts
PowerSum = 60.6Watts.

FSS:
It has been locked for 0 days 0 hr and 23 min (should be days/weeks)
TPD[V] = 1.705V (min 0.9V)

ISS:
The diffracted power is around 1.5%
Last saturation event was 1 days 19 hours and 33 minutes ago (should be days/weeks)
 

WP7845 h1sqz model change and DAQ restart.

Following a restart of the h1sqz model (remove 1 slow, 1 fast, add 1 fast channel to DAQ) the DAQ was restarted.

2018_10_01 13:45 h1sqz
2018_10_01 15:08 h1broadcast0
2018_10_01 15:08 h1dc0
2018_10_01 15:08 h1fw0
2018_10_01 15:08 h1fw2
2018_10_01 15:08 h1nds0
2018_10_01 15:08 h1nds1
2018_10_01 15:08 h1tw0
2018_10_01 15:08 h1tw1

• No initial alignment, IFO locked nicely.
• Ran A2L to see where the spots are starting at. (~6:30am)
• Took a quick CSOFT P OLG meas.  Not great coherence, but looks the same as it has, with the wrong phase just below 1 Hz. (Gabriele said that to run the measurement, you have to engage the higher gain loop)
• Moved SOFT offsets and POP_A offset to try to minimize pitch dither lines on test masses.  I think I'm better centered on IX and EY, but about where I started on EX and IY.  There was a place where I could get EX centering better, but that made the Yarm optics much less centered.  Recycling gain after this is about 49 rather than 50.  CSOFT_P OLG looks the same.
• Running A2L at 8:15am.
• Gabriele moved CARM-type offsets.  None of the offset sliders on the common mode board seemed to have an effect.  Putting an offset in LSC-MCL we could make the arm powers go down.  +/-3000 count offset was okay, and could see an effect.  Max arm power wasn't centered around 0 counts offset.  Trying -4000 counts offset caused lockloss.

Sheila, Haocun, Terry, Nutsinee

We moved the PSL feed forward summing point so that it happens before the OPO frequency servo filter bank.

WP#7845

All T240 proof masses within range, 1 STS proof mass out of range. Relevant script output:

All T240s proof masses that within healthy range (< 0.3 [V]). Great!

There are 1 STS proof masses out of range ( > 2.0 [V] )!
STS A DOF X/U = -8.06 [V]
All other proof masses are within range ( < 2.0 [V] )

It appears we are "steady as she goes".

PR3 YAW displaced by about 20 urad, ETMY YAW approaching 10 urads.

This closes FAMIS 11185

-PR3 Oplev alignment

-Tuesday Maintenance Work (Laser Safe 8-10 am):

• Tumbleweed clearing inside X-arm
• ASC Model update
• OAF Model update
• VE - OMC tube port bag
• VE - MX AIP repair, crane
• EX Pcal work
• Quarterly DUSTMON check
• Beckhoff update
• Beckhoff hardware investigation

   Checked both filters and chillers. 

   All is in order. Both filters are clean and clear, except as noted in previous reports there is yellowing in the crystal chiller filter due to plumbing work on the crystal chiller circuit. No debris was noted in either filter canister.  

   Did not any cooling water to either chiller. 

Closing FAMIS ticket #8313. 

It looks like the optimal buildup happens for a PRCL offset slightly away from zero, i.e. -100

Not much got done. I installed a LP with Fc = 800Hz and -20dB attenuation between the output of LO common mode board and input of PZT driver. Because we can't bypass the PZT driver (Beckhoff needs it to talk to in-vac driver chassis) I turned on the 4Hz/400Hz compensation. OUT2 on the LO common mode board also get signal from IN2 and hooked up to IN1 so we can use both gain slide bars to give it even less gain. I can close the loop with the lowest gain setting without killing the OPO lock but non of the sign combinations (at input and slow) would lock the loop. Looking at IMON on SR785 the noise was pretty much all over the place but ~50kHz-70kHz seems to be its favorite hang-out spot.  I *think* we need a lot of gain to suppress high frequency noise but also can't have that because it will kick the OPO out of place (?). Right now the error signal comes out of IQ demod box. Will it help to install a PFD?
 

Feed forward work: I played around a little bit. Funny that I can't seem to make a different whether I get the output gain to 1, 100, or 1000. Even giving it an arbitrary offset wouldn't make a different to VCO frequency difference. The feed forward signal did show up on H1:SQZ-OPO_FREQ_OUTPUT (summing point according to the front end model).

Today I also learned that open/close the table can make >10mW difference in the Mephisto output power. I couldn't engage the EOM path until the power was stabilized.

Attachments: 1) LO config I've tried. 2) Spectrum of LO IMON

One possible suspect for the CSOFT problem is that one of the test mass actuators did not behave as expected. I injected noise directly into the L2_P stage, and measured the transfer function to DHARD_P error signal (I used DHARD since it is the most sensitive error signals).

All test masses L2 actuators behave as expected in pitch.

Spoiler alert: no success (yet) in fixing the right-half-plane zero in CSOFT_P

Attempt 1

I added a line at 1 Hz on CHARD_P (amplitude 1000 cts) and could see it well in both CHARD_P and CSOFT_P. The transfer function at the line was CSOFT_P / CHARD_P = -1.3e-5, although the phase was not exactly 180 degrees. Nevertheless, I used this value of 1.3e-5 in the CSOFT blend filter bank, and could reduce the CHARD_P to CSOFT_P coupling by a factor of 3-5 or so (compare green trace TF with blue trace TF).

I measured again the CSOFT plant and found that the right-half-plane zero is still there.

Attempt 2

Since I saw a different optimal gain at different frequencies, and some phase rotation, I set up a sweep sine measurement of the coupling. The result is shown below, for frequencies between 0.5 and 2 Hz. There is indeed some structure and some (not too small) phase rotation, ecpecially around 1 Hz. The blue dot are the measurements, and the orange trace is a fit:

I implemented this filter in the CSOFT blend filter bank and measured again the coupling with the subtraction on. At all frequencies the coupling is now at least 20 times smaller. With a frequency dependent constant I was never able to get better than a factor of a few at frequencies around 1 Hz:

Unfortunately, this has almost no effect on the bad zero in the CSOFT_P transfer function.

Stefan, TVo, TJ, Danny

      Stefan has the idea of significantly reducing the amount of time it takes a test mass to reach a final lens by utilizing the ring heater step response and carefully designing an inverse filter to realize the inverse response.  We tested the idea in a previous post using Comsol data and were able to get a time series of the ring heater settings necessary to achieve a "step lens". To test this with measured data I took a time series  from H1:TCS-ITMY_HWS_PROBE_SPHERICAL_POWER after the ring heater settings had been changed by a reasonably large amount and attempted to perform the same procedure. The matlab zpk model of the system filter and inverse filter are attached figures. There are also comparison figures of the ring heater inputs and another figure their respective responses. 

Testing the new ring heater time series: 

With an initial lens around 3 microdiopters, the following figure shows convergence to a final lens of approximately -16 +/- 3 microdiopters within the first three hours but is about 1/2 the predicted lens. I might need to take another look at my model to see why this might be. 

I cut the script the short because there will probably be commissioners coming in tomorrow so I am using this method (on workstation ZOTWS2) in attempts to restore the original ITMY lens by tomorrow morning (but will run till 9/31 00:00 PST to maintain the lens). 

Haocun, Nutsinee

Almost every locking loop in SQZ land is now automated. I haven't created a manager node yet so you'll have to lock each loop by hand. Starting from SHG>OPO>PLL>FREQ>CLF (necessary for working on 3MHz locking loop).

SQZ_OPO Guardian will feed beat note error signal to laser temperature when it's in DOWN and feed PZT offset to pump laser temperature when it's LOCKED_ON_DUAL (or LOCKED). It will also disengage EOM when it's rail and stay in the CHECK_EOM state until somebody close it by hand (to make EOM relock itself requires a little more thinking. It probably could look at PLL guardian state).

SQZ_PLL (phase locking loop) will take care of beat note locking using OPO PZT once the OPO is locked (with TTFSS). SQZ_FREQ will feed (frequency) error signal from OPO common mode board fast path to OPO PZT once SQZ_PLL is locked.

When SQZ_OPO is down it will also take PLL, FREQ, and CLF down with it. Once OPO is relocked everything down the path still has to be locked by hand. PLL will complain if OPO isn't nominal and FREQ will complain if PLL is not nominal.

Neither feed forward nor 80MHz VCO is touched by Guardian yet. Everything worked repeatedly at least all day today. Will try to leave everything locked overnight.

All the SQZ guardians window can be found under SQZ GUARDIAN OVERVIEW button. I will create a compact overview of these guardian windows on SQZ_OVERVIEW later.

Terry also made a nice diagram of our current locking scheme. See attachment. Hopefully it'll be in the dcc soon.

Next: Lock the LO.

Summary

I think Hang's conclusions about the sign flip in the sensing matrix for CSOFT are wrong. Here's why, in brief:

1. the new sensing matrix (with hang's changed sign) has degenerate rows for DSOFT_P and CSOFT_P, so with that configuration we are sensing only one of the two degrees of freedom
2. I tried to step CSOFT_P and DSOFT_P (with a step of 0.3 slider units, using Hang's script move_ARM.py) and found that both CSOFT_P and DSOFT_P signals are sensitive ONLY to DSOFT (see plot below)
3. I measured a sensing matrix at DC, and found that the signs do not match what reported by Hang's measurement.
4. Inverting my DC measurement of the sensing matrix and using it, I get a nice decoupling of CSOFT_P and DSOFT_P.

Details

Step responses

Below the response to a step (with all soft loops open) in both CSOFT and DSOFT, by 0.3 slider units, using Hang's move_ARM.py script. The plot right below shows the response of both DSOFT and CSOFT error signals with the sensing matrix that has Hang's change of sign. Clearly both signals respond only to DSOFT, and are orthogonal to CSOFT. So even though in this configuration we are closing both CSOFT_P and DSOFT_P, we are controlling only DSOFT_P.

Sensing matrix at DC

I measured again the CSOFT / DSOFT / TMSX / TMSY sensing matrices for the four TRANS QPDs, using the same script used in the past. The results for pitch and yaw are reported below, in units of QPD signals over test mass and TMS slider values:

Comparing the sensing matrix elements for CSOFT in my measurment and Hang's measurements, there's a sign flip for all Y QPDs. I am not sure what the origin of this difference is. However, I inverted the pitch and yaw matrices, to obtain the following:

Based on those measurements, and rescaling the elements so to match the size and sign of the current, error signals, I computed a new ASC input matrix for soft pitch and yaw and implemented it. It's very close to the old one, before Hang's sign flip.

New step response

With the new input matrix, the CSOFT and DSOFT error signals behave as expected:

To cross check, I also stepped CSOFT_Y and DSOFT_Y, and foud the error signal properly decoupled:

A also did a quick measurement, with both CSOFT and DSOFT PIT loops closed, by injecting a line at 5.1 Hz. When injecting a DSOFT line, only the DSOFT signal sees it. Instead, when injecting a CSOFT line, there is some cross-coupling to DSOFT: the DSOFT signal also sees a line at 5.1 Hz, just a bit smaller than what is seen in CSOFT. It's not clear to me why this is the case. See fourth attachment.

I tried to bag and then evacuate the leaking 12" CFF flange so at to reduce the leak rate until this can be repaired in NOV. but my attempt was thwarted by the unique fabrication of this 10" nozzle.  As Gerardo M. pointed out to me later, the nozzle roll has a vacuum weld on the interior of the roll seam and an external stitch weld.  This is unexpected as all "normal" port nozzles utilize full penetration seam welds.  The net effect of this is that a channel is left on the external side of the seam and this channel makes it impossible to seal a bag against (without cheating and using some sort of fluid sealer).  I'll point this unique detail out to Chandra R. next week and she can decide if the benefit of potentially reducing the leak rate is worth the use of :the required fluid sealant.