Assembly/In-vacuum cabling - Hugh, Jim, Greg, Hugo
ready
It appears that communication was lost to the end X weather station at around 16:44 7/27/2012 UTC. Restarting the IOC did not re-establish communication. I have not driven out to the end station to investigate further yet.
J. Kissel, D. Barker Recently, the ${userapps}/release/sus/common/models/SIXOSEM_F_STAGE_MASTER.mdl had been updated in order to include signal exchange and CART2EUL and EUL2CART transformation with the ISI signals. Because the TMTS models use this library part, the ${userapps}/release/sus/common/models/TMTS_MASTER.mdl main library part, and the top level model, ${userapps}/release/sus/h2/models/h2sustmsy.mdl needed to be updated to handle these new input and output ports. I've done so on both the above mentioned models, and recompiled to ensure functionality (and was successful). Since this is an unimportant change, at the advice of Alberto, Dave and I agreed to just reinstall, restart, and restore the model during next Tuesday's (July 31 2012) maintenance period. Further Notes: --------------- - the original motivation and discovery of the problem was because Dave has noticed that the /opt/rtcds/lho/h2/chans/ipc/H2.ipc (the file that stores the channel mapping for inter-process communication) had a whole bunch of old crap in it, so he deleted it, and wants to recompile all models that use IPC in order to refill it. - These changes were made in parallel with other changes to top level models. Unfortunately, the commit log message that was intended for the h2susitmy commit - I've also added EPICs output channels which capture the IPC receive errors, called H2:SUS-TMSY_ISC_IPC_PIT_ERR H2:SUS-TMSY_ISC_IPC_YAW_ERR which eventually can/should be included somewhere in MEDM screens.
I've added the following channels to the framebuilder: $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG1_IN1 2048 $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG2_IN1 2048 $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG3_IN1 2048 $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG4_IN1 2048 which required adding these channels to the ${userapps}/release/sus/common/models/QUAD_MASTER.mdl then recompiling, reinstalling, restarting, and restoring the QUAD test mass models: ${userapps}/release/sus/h2/models/h2susitmy.mdl ${userapps}/release/sus/h2/models/h2susetmy.mdl Before restarting, I created new safe.snap files, in /opt/rtcds/userapps/release/sus/h2/?tmy/h2sus?tmy_safe.snap in order to capture the new optical lever calibration. After restarting, I restored to the following snap files: /ligo/cds/lho/h2/burt/2012/07/27/14:00/h2sus?tmyepics.snap
While bringing things back up on ITMY, we noticed rather large motions around the BSC8 chamber, as noted in Thomas' LHO aLOG 3634. We were concerned that it had to do with the restart of the model (a la March 2012), but as Thomas shows the large motions had started to occur prior to our model restarts. Bram also reports that the cavity had been behaving poorly for an hour or so prior to this computer work. Finally, because we were scared, we manually triggered the entire chambers IOP watchdogs, so we were confident it was not any excitations from any DAC. We further confirmed this by checking both the SUS coil-driver monitor signals, and noted that the ISI's "rogue excitation" monitor also remained green. The large motion dissipated after ~ 5 minutes of just waiting (we would have turned everything back on, with damping loops running -- which is more safe that with everything turned off, if we trust CDS infrastructure to be functional -- but the the GS-13s were saturating, which would not allow ISI watchdogs to be reset). It should be noted that though these motions were large, they were not extreme. The QUAD's top-stage OSEMs were visibly wobbling, but only by a few thousand counts. ISI GS13s were railing, but they're extrememly sensitive, and ISI position sensors on both stages indicated that were not hitting the locker-limiter mechanical range of motion. Also worthy of note -- the recompile, reinstall, restart, and restore of H2 SUS ETMY went smoothly, without trouble. Because adding the channels to the DAQ list is part of the QUAD_MASTER library part, the change was literally identical in both models.
Before we started the fine tune of the balance for ISI Testing, I locked up the HEPI. There appears to be a real rotation of the HEPI as measured by the Dial Indicators compared to the signed off position of Wednesday 7-25. I locked down corner 1 (NE Pier) and then locked up Corner 4 (SE). When doing this, one watches the dial indicators keeping the Indicators reading the recorded position. When I went to corner 3 (SW), I noticed a big offset. I was suspicious of the cable pulling activity had disturbed this DI. So I released all the stops again and I see a coherent shift in the indicator readings. This suggest to me that it wasn't any DI bump but something real. If so the move is on order of 0.5mm ==> likely about 2x requirement in rotation. I was rushed a bit hoping to get the ISI ready for weekend testing so I just locked the HEPI. When testing is complete I want to do a more careful job of unlocking HEPI and make sure nothing is pulling it out of position. If it still looks real, I may request another IAS look to confirm or correct what I see.
- Cavity Scanning: Daniel - H1 SEI chassis powered down - H1 DAC restart - H2, added ISC and slow EPICS channels to front end - Large oscillations on ITM ISI and quad. Watch dogs tripped at about 22:47:50 Locked doors at 4:36 PM Pacific time
HughR, HugoP, GregG, JimW This afternoon HAM3 was finally fully balanced and the Stage1 left floating for testing over the weekend, prior to next weeks planned install activities. It's condition is a little rough, with one locker that seems to need a shim adjustment and a number of SEI-responsible SUS/ISC/what-have-you cable-ends that are sitting on the optical table because they have nowhere to go yet (they are, however, clamped to the table to keep them from rattling too much, see pic). There are also some small issues with the lock/unlock positions that maybe due to temperature differences or differences in support structure (staging building test stand vs HEPI/crossbeams/etc) that we won't fix until the payload is in a more final state.
The PR2 now in chamber-side testing has had its matrix and filter bank settings installed. A good snapshot to revert the PR2 lives in: "/opt/rtcds/userapps/release/sus/h1/burtfiles/20120726_h1suspr2.snap"
J. Kissel, J. Garcia, D. Barker
The Simulink user models were modified today to incorporate IPC functionality on 'h1sush34'
. This library part is using Shared Memory on "h1sush34" only, so communication across different I/O chassis is not needed. The motivation behind the modifications was the fact that the BIO switch cards allow only one user model to send and receive signals to the hardware. So, the solution was to use the MC2 user model, "h1susmc2", as the communicator to the BIO card hardware. Signals sent and received to the BIO card must be routed through the MC2 user model. The control input, however, is controlled in the individual models ("h1suspr2" & "h1sussr2") but sent to MC2 by shared memory to switch the BIO card hardware. A signal to the BIO card for the PR2 channels can still be entered from a PR2 medm, but in reality the signal is routed through the MC2 user model. So, in order to control the BIO switched for PR2 or SR2, the "h1susmc2" model must be running for the full functionality.
All three models were recompiled and installed on "h1sush34" along with a reboot of the DAQ.
Models are checked into the svn locally in:
/opt/rtcds/lho/h1/userapps/release/sus/h1/models/
"h1susmc2"
"h1suspr2"
"h1sussr2"
An issue arose today after installing the PR2 model in that the drive signal from a DTT session would not begin in the user model channels. As it turned out to be, the DCUID number used in the PR2 model (#40) was used previously by a PEM model no longer in use. The DTT issue arose when the par file for the testpoint manager had the DCUID number assigned to the previous PEM model. This par file has been edited manually to remove the previous DCUID and replaced with the PR2 DCUID. Excitations could then commence on PR2.
(CHANNEL, OLV, Offset, Gain)
TOP BOSEMS M1
H1:SUS-PR2_M1_OSEMINF_T1_INMON 29963
H1:SUS-PR2_M1_OSEMINF_T2_INMON 27895
H1:SUS-PR2_M1_OSEMINF_T3_INMON 28003
H1:SUS-PR2_M1_OSEMINF_LF_INMON 28986
H1:SUS-PR2_M1_OSEMINF_RT_INMON 30722
H1:SUS-PR2_M1_OSEMINF_SD_INMON 27004
MIDDLE AOSEMS M2
H1:SUS-PR2_M2_OSEMINF_UL_INMON 25961
H1:SUS-PR2_M2_OSEMINF_LL_INMON 28699
H1:SUS-PR2_M2_OSEMINF_UR_INMON 25278
H1:SUS-PR2_M2_OSEMINF_LR_INMON 25061
LOWER AOSEMS M3
H1:SUS-PR2_M3_OSEMINF_UL_INMON 23599
H1:SUS-PR2_M3_OSEMINF_LL_INMON 24880
H1:SUS-PR2_M3_OSEMINF_UR_INMON 26085
H1:SUS-PR2_M3_OSEMINF_LR_INMON 24842
From the above OLVs, offsets were calculated -(x/2) and gains were calculated 30,000/x, snapshots attached.
Travis set the 6 Tops to 50% OLV in prep for Phase 2a TFs.
This is an update to the calibration performed on June 25 2012. The TMS offsets have changed significantly since the previous calibration. Also note that the channel name for PD1 has changed from CHAN_30 to CHAN_26.
PD3 = CHAN_28 (P,Y) = (-10133, -7649) |
|
PD1 = CHAN_26 (P,Y) = (-79733, 7371) |
PD2 = CHAN_29 (P,Y) = (-79933, -7609) |
PD4 = CHAN_27 (P,Y) = (-42586, -7448) |
For PIT, PD3 and PD2 are separated by about 6.25 in and 21600 counts. The corresponding calibration is 6.25*25.4e-3/4000/21600 = 1.84 nrad/ct.
Also for PIT, PD4 and PD4 are separated by about 11 in and 37100 counts. The corresponding calibration is 11*25.4e-3/4000/37100 = 1.88 nrad/ct.
For YAW, PD2 and PD1 are separated by about 11 in and 14980 counts. The corresponding calibration is 11*25.4e-3/4000/14980 = 4.66 nrad/ct.
Range is calculated using +-131072 cts in the TMS pointing range.
PIT | YAW | |
Calibration |
1.86 nrad/ct Positive pitch lowers the beam. |
4.66 nrad/ct Positive pitch moves the beam to the left (PD2 > PD1 direction). |
Range | +-0.243e-3 rad | +-0.610e-3 rad |
The ISC signals have been added to the ETMY HEPI model. Changes have been committed to the svn. The model has been successfully compiled and installed. We should be ready to offload.
To offload to HEPI we need to low pass the cavity signal before to add it to the HEPI longitudinal super-sensor. We made preliminary measurements this morning to check the Cavity and HEPI signs and units, but we need to refine the drive parameters. The template is saved in: /ligo/svncommon/SeiSVN/seismic/HEPI/H2/ETMY/Misc/DTT_Cavity_offload What we need to do next week: - adjust the measurement parameters to get good coherence - tune the gains and check the signs to put the signals in the same unit - design the low pass filters. We plan to make a very low frequency one, to reject tidal motion. And a higher frequency one to reject the motion amplification due to ISI gain peaking (around 30 mHz) - measure the open loops (ISC, HEPI, and the sum of the two)
J. Kissel, T. Vo We received the daughter boards (D1001631) which take place of the Binary I/O connection for control of the Oplev Whitening chassis (D1100013), specifically the internal ISC Whitening board (D1001530). For each of the 4 channels, the whitening board has 7 independently switchable states: (1) 24 dB (x16) Gain (2) 12 dB (x4) Gain (3) 6 dB (x2) Gain (4) 3 dB (x1.4) Gain (5) [10:1] Whitening (6) [10:1] Whitening (7) [10:1] Whitening where my notation for zeros and poles is [z:p]. So, Thomas and I installed the daughter on H2 SUS ITMY's optical lever, and explored the raw voltage spectra of the QPD segments with each of the settings on (well, most of them). I attach the results. To fill out the legend: RED: This was the state that the optical levers had been in, with no control (i.e. the BIO input spigot was left open). BLUE: Daughter board installed, but with all switched flipped to OFF. This is the same as red -- good. That means the chassis doesn't *need* the daughter board to be functional, and the board sits at its default state (no stages engaged) with nothing plugged in. Fair enough. GREEN: 12 dB Gain stage turned ON BROWN: 4 dB Gain stage turned ON MAGNENTA: One [10:1] whitening stage turned ON CYAN: Two [10:1] whitening stages turned ON BLACK: All Three [10:1] whitening stages turned ON Note that we tried switching on the 24 dB gain stage (alone), as well as having both 6 dB gain + One [10:1] whitening stage. Both of these configurations saturated the ADC. Assuming the amplitude of the signal represents a reasonably quiet time, we decided to leave the chassis in the CYAN state, with two [10:1] whitening stages and no gain, to give some head room for noisy day time activity, or when the ISI is not functioning as well as it is now. Because we now have whitening in the chain, these filters have been compensated for in the $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG# banks (in addition to the -1 discussed in earlier logs). One thing to note: the ADC noise floor (visible in the unwhitened spectra, above ~40 Hz) is 5 uV/rtHz. Calibration Details: -------------------- Here's a schematic of the QPD signal chain: +--- $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG1_IN1 +-------+ | Up ^ | 1 | 3 | +-------------------+ +-----------------------+ +-----+ +--- $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG2_IN1 | |---+---| ---- | Whitening Chassis | ---- | Anti-aliasing Chassis | ---- | ADC | ---| | | 4 | 2 | +-------------------+ ^ +-----------------------+ +-----+ +--- $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG3_IN1 +---+---+ | | | +--- $(IFO):SUS-$(OPTIC)_L3_OPLEV_SEG4_IN1 equivalent voltage measured here (staring at the face of QPD, as though you were an incident laser beam). We used DTT to measure the above listed channels, which were then calibrated using the 16-bit ADC calibration of 40 / 2^16 [V/ct] = 6.1e-4 [V/ct] hence, as I point out in the diagram, we're effectively measuring the output of the whitening chassis since the AA chassis has a gain of 1 and no frequency response.
J. Kissel, T. Vo We performed the exact same measurements on ETMY this morning. The results were virtually identical, so we left the ETMY Oplev Whitening chassis with two [10:1] whitening stages engaged. I've compensated these stages digital in the SEG filter banks. Let's measure some real motion!! For the record, the .xmls that take these measurements can be found here: ${SusSVN}/sus/trunk/Common/Misc/ 2012-07-26_H2SUSITMY_OplevWhiteningStudies.xml 2012-07-27_H2SUSETMY_OplevWhiteningStudies.xml
For the record (and long overdue), here is the payload weight of the SUS components on BSC8:
44 lbs = 8 x 5.5 lbs for Vibration Absorbers
~60 lbs = FMy Stays
~2 lbs = FMy Sheer Plate dampers
10 lbs = Cabling
2 lbs = Cable Brackets
3 lbs = Ring Heater + Brackets + Cables
266 lbs = ITMy QUAD Upper Structure
531 lbs = ITMy QUAD Lower Structure
44 lbs = QUAD Sleeve
2 lbs = QUAD sleeve wedges
341 lbs = FMy SUS (weighed in at 394lbs with the 53 lb LSAT attached prior to install)
55 lbs = 1.26 lbs X Dog Clamps (33 long + 22 short)
TOTAL = 1360 lbs = 641 kg
The Arm locks again, with the spectra attached (first attachement). The positive message is that the ISI performance has greatly improved. The drop at 0.5 Hz is the Quad pendulum mode. At higher frequencies (above ~0.6 Hz) we are most likely dominated by frequency noise.
We optimised the SHG temperature, so there is more power in the green (an extra 10mW). This is meant to remidy the power fluctuations we were seeing when we drifted further away from our nominal VCO frequency. We will see if that is the case. I reduced the CMB-B input gain by 3dB (down to -17dB!) to make up for the increase of light on the diode, but I forgot to measure the UGF of the PDH loop (nominal 11 kHz).
Local damping to M0 and L1 on both Quads are engaged, and the length feedback is of loaded to the ETM M0. The power drops over the longer time scale (>1 hr), is most likeli becasue I haven't inserted the decoupling. I tried to decouple it in pitch without much succes.
In the spectrum, the features at 2.75 Hz and 4 Hz are seen in the Reference Cavity transmitted power, and are imposed on the laser in the end station.
The problem we currently having is that the VCO seems to jump every ~8 min, when it reaches 40.2 MHz (about 600 kHz from the nominal 39.6 MHz), seen in the second attachment. The cavity seems to stay lock and all keeps running 'smoothly'. Due to this jump I was not able to do more than 4 averages in the spectrum.
The cavity has been locked for over an hour, which is an improvement. I unlocked the cavity before I left.
On request ... the traces are taken at 1) 27/7/2012-02:47:57, 2) 19/7/2012-7:17:00, 3) 19/7/2012-6:40:00 and 4) 19/7/2012-7:40.
For funzies: comparing RED trace against Figure 3 of T1100080. Pretty darn close to what was expected!