Heard the chiller alarm this morning.  Found the water level to be a little below minimum.  Added about 400 milliliters.

model restarts logged for Sat 13/Sep/2014
2014_09_13 06:27 h1fw1
2014_09_13 09:52 h1fw1

unexpected restarts of h1fw1

Today I spent some time working on the DRMI guardian. It now is controling the state of the ITMX, IMTY, PRM and SRM suspensions.  It can transition smoothly between PRX, PRY, and MICH on the dark fringe with just one click. Hopefully this makes the repeated alingments of PRMI easier.  

Notes:

• Due to the buggy input matrix loading, you might need to check that the elements that got set are actually set (look for red).  The guardian is checking for this error in PRX PRY, but I didn't get to addding much error checking to the MICH or PRMI states yet. 
• None of these states can cope well with the situation where something gets kicked and is saturating, if that happens human intervention is needed, you should be able to simply request the down state, wait until everything is settled down, and then request the state you want again. 
• You will have to save alingments before changing the guardian requested state to a different configuration, otherwise your alingments will be lost.  The DRMI guardian should warn you with a user message i this is happening. 
• I made only the states which are tested requestable, so if it is on the pull down menu I think that it should be working.  The only questionable one is PRMI carrier, which was working OK last night but I didn't have time to actually see it lock today.  
• When one of these configurations losses lock, they do not go back to the down state (unless the IMC also lost lock), they just go to the LOCK state.  If you want to start from scratch and reset parameters, manually select down.  

model restarts logged for Fri 12/Sep/2014

2014_09_12 12:52 h1iopseih45
2014_09_12 12:54 h1hpiham4
2014_09_12 12:54 h1hpiham5
2014_09_12 12:54 h1isiham4
2014_09_12 12:54 h1isiham5
2014_09_12 16:22 h1broadcast0
2014_09_12 16:22 h1dc0
2014_09_12 16:22 h1fw0
2014_09_12 16:22 h1fw1
2014_09_12 16:22 h1nds0
2014_09_12 16:27 h1nds1

no unexpected restarts. h1seih45 restarted to clear stuck DACs. DAQ restarted to synchronize h1omc.

This morning at 2:30 am pacific time ITMY controls were turned off to check that the sawtooth seen on the OpLev yaw is not somehow cause by the ISI.  The sawtooth was there all night, and not seen in the stage 2 Rz sensors, screen shot attached.

This is as expected, since if this were real motion of the suspension we would not be able to lock. 

The ISI tripped this morning and tripped again when I was trying to bring it back, with the guardian complaining about smoe filter settings not as expected. 

Now it is fully isolated.

That's 1237pdt, I can't see Jim's matlab sesssion with the locked out terminal.

Loaded these this morning after getting the new .mat files Hugo built yesterday.  A few minor plotting difficulties and then after conferring with JeffK, loaded the filters at 1228pdt.  Opened the HEPI loops, loaded the filters and turned them back on all with the ISI Isolating (Lvl1.)  Silly me, I didn't take a DDT look before the switch altthough I had done this before a week+ ago and I could see the ugly zero pretty easily then; the second plot in 13810 is a good place to see this.  I undamped/unisolated the ISI and looked with DTT and like the long commissioning transfer functions, you really can't get any coherence in the area but at least the amplitude doesn't show the deep zero.  Next step would be to run the longer TF on the HAM4 ISI--later.

Alexa, Rana, Sheila, Kiwamu

We continued working on the carrier PRMI tonight as a part of the too-low recycling gain study. No new result -- the recycling gain stayed as high as 15 and reflection remained 30-ish %.

Then we switched the effort to the sideband locking to get ready for the upcoming full interferometer locking, but no luck in catching a fringe for unknown reason. We woll continue working on the sideband locking during the weekend.

(Carrier lock automation)

We worked on an automation of the carrier lock using the guardian for the carrier lock configuration. We tested the code by letting it lock the PRMI. It runs fine.

(Camera adjustment on PRM, PR2 and PR3)

As a part of the intracavity clipping study, we readjusted PRM, PR2 and PR3 GigE cameras to have a better focus on the infrared images. The attached are screenshots of those cameras after the readjustment of focusing and zooming. The PRMI was locked on the carrier.

Some observations:

• PRM baffle tends to show a scattering on the left side of its image.
• PR2 image showed a small spot-like scattering. We did not locate where it correponds to.
• PR3 tends to have a bright scattering below the optic even though the beam does not look super low on the mirror.

(PRMI carrier WFSs preparation)

We started looking at the WFS signals in order to aling the PRMI without keep clicking the sliders on the screen. Tonight, we made a single measurement -- which is a sensing coefficient from IM4 to WFS A and B.

WFS_A_RF9_I_PIT = 4.8 [wfs cnts/ IM4 pitch cnts]  (-172 deg)

WFS_B_RF9_I_PIT = 6.6 [wfs cnts/ IM4 pitch cnts]  (1 deg)

WFS_A_RF9_I_YAW  = 0.9 [wfs cnts/ IM4 pitch cnts]  (0.4 deg)

WFS_B_RF9_I _YAW = 1.36 [wfs cnts/ IM4 pitch cnts]  (3.5 deg)

Fully automated so only watchdog trips and an undamped SUS would spoil things.  When the TF is done, the Guardian will be unpaused to bring things back to nominal.

Here's some plots showing the kind of performance we have in the corner station. Nothing new here, just wanted to do a quick status update.

For the 3 chambers, we can make some general comments:

- we're reinjecteing some noise below 100mHz in X and Y

- we're hitting the noise floor pretty quickly in the rotational degrees of freedom.

- some isolation is provided by stage2, but it's not very aggressive (lot of room between 1 and 10Hz). The good thing with this configuration is that we're not introducing some extra noise at low frequency (this comment is true for ITMX and ITMY. BS is on damping only for ST2).

Jason, Rana

Rana had the idea that the ITMy problem was due to the thermal electric cooler not effectively cooling the laser therefore causing the instability we are seeing as a sawtooth yaw signal.  To test this we first increased the laser power and then decreased the laser power, both times looking to see if the sawtooth changed at all.  In the attached SUM graph:

• The first jump is when we turned the laser on after closing up the receiver box
• The second jump is Rana turning the laser power up
• The third and fourth jumps are removing the Lexan viewport cover and re-aligning the oplev
• The drop at the end is me lowering the laser power

Throughout this it can be seen that the sawtooth pattern doesn't change, so that rules that idea out.  Will continue to continue to investigate...

We've reduced the motion on PR3, especially the 'bump' seen in the optical lever pitch around 0.7Hz.

Turns out there is a good coherence between GS13 X and the optical lever in pitch. The GS13 is far away from the noise floor around this frequency, so by lowering the blend frequency in X, we were able to gain a factor of a few without introducing too much gain peaking at lower frequencies.

Also, by switching the GS13s to high gain, we improved a little our performance at low frequency in RZ/Yaw.

This configuration seems good enough for now, but the next step is to design a more fitted blend set for HAM2. I'll add that to my to-do list for next week.

Attached are the TF taken Thursday morning.  Like those seen on HAMs 4 & 5 ISI, having the HEPI unlocked and position loops closed versus HEPI on stops, the HEPI structural bumps between 10 to 40hz, are reduced in frequency and lowered in Q, see the plots in 13763 for this comparison.  Unlike HAMs 4 & 5 however, which have generic 5hz HEPI loops, the HAM3 ISI doesn't see the H3 dip at ~6.5hz develop into a nasty high q zero; the HAM3 HEPI horizontal DoF controllers are 2hz ugf non-generic.

Hugo has produced new generic controllers with the 2hz ugf, I'll upload those as soon as the machine is available.

Restarted data concentrator at 16:20:50 to clear the bad DAQ status for the h1omc model which was modified Sept. 9.  

The h1nds1 daqd process did not restart itself, and had to be manually started with monit.

08:00 (carryover) LVEA is LASER HAZARD

09:30 Cris at -End-X

09:59 Kyle at End-X

10:14 Karen at End-Y

10:29 Reset HEPI watchdog accumulators for BS and ITMX. This had not been done as part of the Tuesday weekly routine.

10:32 Kyle back from End-X

10:33 Jason out o ITMY OpLev

10:36 Karen leaving end-Y

10:41 Travis out to LVEA to worlk on Quad Test Stand

11:50 Kyle out by HAM6

12:43 HEPI end-Y threw an alarm for pump differential pressure

12:49 Jim to restart Seismic models on HAMs 4 and 5

14:24 Corey out to East bay to look for parts

14:50 Corey out of LVEA

15:16 M Landry into LVEA to walk around in prep for Saturday Tour

The DAC enable status of the h1iopseih45 model went bad at 14:16:44 PDT on Sept. 11.  This required all models to be killed, then the IOP model to be restarted, followed by starting all the user models again.  At the time the DAC enable went bad, the IOP state word also reported a FE, Timing, and ADC error, which were cleared leaving the DAC enable error which requires a model restart to clear.

Centering: Done.

I continued the POP sled centering this morning and now it's good (see yesterday's entry on the same subject: https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=13890).

I ended up moving both PIT and YAW by large amounts. Though the picomotor count is not realiable as a reference (because it has hysterisis and because the software sometimes skips the counting for some reason), just to see how much I moved, I show those numbers anyway.

With the whitening gain of 45dB, QPDA and QPDB SUM (raw counts) were 1595 and 1463 counts on average for a straight shot beam, while we expect 1540 for both assuming 10W through IMC (it should be more like 8W in reality), so the measurement is somewhat larger than expectation but is still quite consistent with expectation.

Whitening gain: Changed to 12dB (from 45dB)

After the above adjustment was done, the whitening gain was lowered by 33dB so that the QPDs wouldn't rail when the PRC is locked to the carrier, and dark offsets were remeasured and set. The nominal whitening gain now is 12dB.

"toW" filters in the input are now "uW"

"toW" filters in FM10 of the input are redone such that the output is the power the diode receives in micro Watts when the whitening gain is 0dB, the filter name was changed to "uW", and I put "-12dB" filter in FM9. This way, when you change the whitening gain all you need to do is to change FM9 or add other fixed gains in FM5-8.

Anyway, I used the following calculation:

Responsivity of the diode is 0.8 A/W. Transimpedance is 1 kOhm (https://dcc.ligo.org/LIGO-D1001974). ADC is 2^15 [counts/volt]. There's no factor of 2 (single ended signal V is sent out as +-V differential, but it's converted back to V at the receiving end).

So the entire chain has 0.8A/W * 1 kOhm * 2^15 cts/V = 2.621E7 cts/W = 26.21 cts/uW.

"uW" filter is just a gain of 1/26.21 = 0.038. (FYI, "toW" filter used to have a gain of 0.0056, don't know why.)

With the straight shot beam, and with "uW" and "-12dB" on while the whitening gain is 12dB, POPA and POPB SUM are 0.34 and 0.31 uW respectively. OTOH 10W(MC)*3%(PRM)*229ppm(PR2)*10%(90:10)*1/2=0.34uW, so it's reasonable.