HAM 4
0905 - Richard working on ETM ESD in control room.
915 - Gerardo + John to MY to relieve pressure on valved out vacuum gauge (currently presenting 6e-5 torr)
925 - Jeff & Andres - Live at the H2 Enclosure one night only for OFI move to HAM5 Area and OFI Install.
945 - Travis & Margot & Fil - ACB staging area in LVEA (test stands)
1045 - Praxair delivery
1102 - KingSoft onsite
1130 - Filiberto climbing on top of HAM 1 & 2
1330 - OFI install halted while Arnaud runs some TFs on HAM5
This morning (9:20am-ish) I turned the purge air back up that is plumbed into the East side of HAM4. It had been reportedly turned down yesterday. Soft covers which were weakly billowing are back to full billow.
I re-ennabled the path from SR2 WD to HAM4-ISI WD in h1sussr2.mdl, then re-compiled and restarted it.
The new model is commited under the svn -r8034
Work was perfomed under WP # 4639 which is now closed.
This morning, I aligned the 3 stages of OSEMS (14 in total). This involved some clean filing of one fo the middle stage brackets. A month ago, we noticed that this bracket held the AOSEM already at an extreme in height range, however after beating the streets to see what could be done (via SUS engineering and SUS LLO), we decided to leave it and hope we didn't need more range. The recent SRM mass alignment has sent a small differential pitch up the chain causing this AOSEM to now need more height range. Dam*. The bracket itself has slots for additional height adjustment, however the bracket then runs into the side of the AOSEM itself (hence the poor design statement above). Since we can't do much at this point to fix the range, I disassembled the bracket, removed the AOSEM and filed the bracket down where it hits the AOSEM. I used a Class B file. There is another AOSEM hugging the edge of the mechanical range, but I'm leaving it for now.
With improved mechanical range on this one AOSEM, I was able to align it and the rest of the OSEMs. I've run quick TFs in V, L, and R - all look good so the chamber is ready for OFI installation (and for Jason to post final alignment numbers of SRM).
Added a ZHL-1A RF amplifier to the 45MHz RF signal which goes to the EOM. The RF power measured at the patch panel in the LVEA is now 25.3dBm from 12.7dBm. There is a 6dB attenuator in front of the RF amplifier which can be replaced with a 12dB one, if the gain is too high. The modulation depth was measured in alog 9395. There seems to be a 2dB loss between the LVEA patch panel and the EOM. If correct, the modulation depth is now 0.3 and the 3f-signal strength about 70 times higher. With a little bit of cable and barrel magic the phase was restored to within 1 degree.
Arnaud and I have made some changes to the SUS guardian. We broke up the graph such that the requestable states are now idle states that don't actually change the state of the suspension:
The dark blue states, which are the requestable states, now only check the alignment and watchdog status. The reason for this change is so that guardian being in e.g. the ALIGNED state actually indicates that suspension is fully aligned. Previously guardian would enter the ALIGNED state, even before the offsets had fully wrapped to their alignment values. Other guardian nodes that might be watching the SUS guardians for optic alignment would therefore be misled. This just makes the state more explict.
I also removed the goto flag from the RESET (was SAFE) state due to a bug I discovered in guardian. The bug is that if there is a goto on the graph, then the INIT state will not be executed. This is something I need to fix in guardian. In the mean time, I just made all the desired transitions explicit.
I wanted to mention a couple more changes that Arnaud and I made to the main SUS guardian module:
Arnaud will write another follow-up comment once these changes are all committed.
The changes were commited under the svn. Details can be found under the OpsManual
Thought the issue was a bad switch, but it was actually one of the cables. I swapped this cable out (which wasn't easy since it was hard to see where the cable connector needed to go [& these connectors suck!]).
This completes WP 4633. This was done to communicate with the MetOne model 6000 installed in the H1 PSL diode room. The h0dustdr, h0dustex, h0dustey, h0dustlab, h0dustlvea, h0dustmx and h0dustmy target directories in /ligo/lho/h0/target were copied to /ligo/lho/h0/target_archive and updated. The h0dustex, h0dustey, h0dustlvea and h0dustlab IOCs were restarted and burtrestored to the snapshot files in /ligo/cds/lho/h0/burt/2014/05/19/08:00. The h0dustdr IOC was started. The settings were changed to: Sample Time: 20 seconds Hold Time: 300 seconds 0.3 micron minor alarm: 100 0.3 micron major alarm: 200 0.5 micron minor alarm: 100 0.5 micron major alarm: 200
Gerardo and I pumped out the dead space around PT246 and closed the valves to their prior state.
SRM looking good; Betsy plans to do final check and hand off to OFI today
Apollo will be craning OFI over beam tube
Gerardo s. closing GV5/GV7 in anticipation of craning
HAM4 ISI floating, HEPI locked
McCarthy and Fil pulling cables at HAM5/6
Dan Hoak/Corey working on SLED buildup for HAM6
TCS calibrating CO2Y, troubleshooting interlock system CO2X
laser hazard expected ~1400
Corey swapping light in ISCT1 and manhandling things with Bubba in the squeezer bay
control room tour expected 1230-1300
King Soft expected
No restarts reported.
Kiwamu, Stefan, Lisa
After talking with our LLO locking friends, spending some more time trying to further reduce the CARM offset, and thanks to the fact that tonight everything else was working fine, we think we know why our CARM offset reduction is so painful.
At LLO, the transition to 3f signals can be done with a very large offset (~ 3kHz), and once the transition to 3f signals is done, there is no particular issue in reducing the CARM offset down; no adjustments are done in the DRMI control.
Here, first of all we can transition to 3f signals for PRMI only after having passed the 18 MHz resonance in the arms (~970 Hz). This implies using the REFL 45 I&Q signal down to 700-800 Hz, which can be done only by tuning the demod phase on-the-fly .
Second, once on 3f, without constantly tuning the MICH loop (gain, offset, demod phase), we can't keep the PRMI locked while reducing the CARM offset.
We realized that the problem is that here we are using REFL 27 Q for controlling MICH, instead of REFL 135 Q. This is bad because REFL 27 Q is much more sensitive to the CARM offset, and offset and gain changes for this signal are indeed predicted by Kiwamu and Anamaria's simulation (though we don't have on hand plots for the exact CARM offsets we are currently at; Kiwamu is in the process of producing more plots).
The reason why historically REFL 135 Q was not used here is that there was essentially no signal. By comparing the LLO parameters with the ones we are currently using, it looks like we should increase the modulation index by at least a factor 2 and increase the light on the BBPD.
On the other hand, once we realized what was going on, we could reduce the CARM offset down to 250 Hz, by keeping the MICH loop happy monitoring offset, gain, and demod phase, and adjust all these parameters by hand. We could probably keep going, but there is no much point in doing that if we believe that we are just making our life harder by using a hostile signal.
Here is the roadmap to reach 250 Hz:
CARM OFFSET / MICH OFFSET
700 Hz / +200 cts
600 Hz / +400 cts
550 Hz / +800 cts
400 Hz / +900 cts
350 Hz / +900 cts
250 Hz / +1300 cts (PRCL offset -1000 cts)
MICH GAIN = 2
PRCL GAIN = 1.5
Other changes we did tonight:
1. Stefan relaigned the full IFO (X arm green, X arm red, TMSY single shot to beat note, Y arm green, PRMI)
2. We added a notch in the PRCL loop for the 70 Hz periscope resonance
3. We realized that theTMSY YAW damping loop has been off since May 15.. we re-closed the damping..
4. We modified the ALS slow path to include a 40 dB 5 second ramp for smooth engaging
5. We re-commissioned the Y arm ITM dither alingment loops, 1 dof only, in PITCH and YAW (the PZT loops are broken for both arms since the WFS software upgrade)
ISS was in a state where he tries to engage the loop but keeps failing. So I re-adjusted the reference DC signal to -1.631. It fixed the issue. The diffracted power is now at 8-ish percent. Good.
Hope this will help our locking activity tonight.
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 :
We tracked down the problems we had in the past windy nights to be related with the ALS tidal tuning , but it is still good to make these plots with the OpLEv signals.
However, I think your plots didn't turn out the way you wanted to. There is only YAW in the second attachment.
Also, it would be easier to make a comparison if you put windy/quiet curves for each DOF in the same plot. Thanks!
thanks for noticing!
Attached are some other plots :
1. wind at corner and endy stations between may 17th/may 20th. High winds references are marked in orange, low winds is in green
2. Spectra comparison of ground sensors (HEPI STS) at end and corner stations with different winds. Blue and pink curves are during high winds (blue = may 17th, pink = may 18th). Red is during low winds (may 19th). From top to bottom rows : ETMX ETMY and ITMY. From left to right : X Y Z degrees of freedom.
3. Spectra comparison of oplevs during high winds (may 18th blue curve) and low winds (may 19th red curve) for ETMx ETMy and ITMy top to bottom. Left column is pitch, right column is yaw.
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.
I closely looked at one of saturday's trip on ETMX @ 04:44:40 UTC (May 18th 2014). The sequence was :
I will post some data tomorrow
This is how the magnet looked in the UR M2 AOSEM before bracket "attitude adjustment".
This is the bracket-AOSEM mechanical intereference that didn't work today. (Note, we've reported that on many of our HSTS suspensions at LHO these brackets are at the end of their range, it just bit us this time. I do not know why they are all high at this stage, but the suspension pendulum modes match the model so the pendulum lengths are correct. Our last round of diagnosis on this many months ago was hand washed off that the LHO suspension structures were possibly from a different batch than the LLO ones and therefore there is a mechanical difference in how these AOSEMs are mounted.)
And this is the bracket with my hand-crafted AOSEM relief shape which shoulda been in the bracket in the first place.