Jonathan Berliner, Rolf Minton, Gerardo Moreno MichaelR (thanks!) tipped me off this morning that Pcal at MX was in the way of the upcoming BSC removal. RickS assigned me to remove the Pcal box. I conferred with DaniA and DougC about locks and tags that were there (none extant). JohnW was concerned about exposing the viewport window. Sure enough, the viewport window did not have a viewport protector (see photo), so I called GerardoM over to cover the exposed viewport. It truly was a 3-person job moving the support structure out due to weight and awkward shapes. New homes for parts: - Support structure - metal disposal by Big Red's parking spot - Clamps, bellows, and assembly - Vacuum shop - Pcal Box, bolts, fasteners, etc. - LSB Optics Lab. We blocked the opening of the box with a dump.
The SUS Team We've found welded repair plugs (by quick visual inspection) in the prism mounting holes of an M2/M3 metal mass (D080368) used in BSFM triple suspensions. See attached photos. Though we replace the lowest stage (M3) of the BSFM's metal mass with fused silica for final in-chamber suspensions, the M2 mass is identical to the M3 mass, and remains in the suspension. In addition, for the H2 OAC, we are to leave FMY's M3 as a metal mass. Two other masses have been checked for similarly visible repairs and none were found. Stay tuned for further details.
B. Bland, J. Kissel, R. Lane, J. O'Dell, N. Robertson, T. Sadecki As was mentioned in Monday's aLOG entry, in order to increase the lowest pitch frequency of the X1 SUS BSFM01, we lowered the M1 Blade Tip heights, such that the measurement between the M1 Base Plate and the Blade Tip break off point is 23.6 mm, (i.e. a d1 = nominal + 3 mm = 4 mm below the COM) as opposed to 26.6 mm ( a d1 = 1 mm below the COM). This *did* increase the pitch frequency, but not as much as predicted by the model -- The pitch frequency moved from 0.44 Hz to 0.46 Hz, where we expect from the model to be at 0.49 Hz. We are still confused as to why this is the case, and the attachment plots several models with the respect to the two measurements, adjusting parameters to try to explain the pitch transfer function. The conclusion is that, without redesign of the mechanical components of the suspension we cannot increase the lowest pitch frequency to more that 0.46 Hz, which is a reasonable 12% above from the first longitudinal mode at 0.41 Hz. We'll consult with other experts as to whether such a deviation from "requirements" is acceptable (though our first impression is that it is). ------------------------ The story (expanded legend) is as follows: BLUE - Nominal Model. This is what we expect the M1 P to P transfer function to look like. Note here, the lowest pitch mode is 0.488 Hz. This is what we *expect* a measurement to show if the blade tip heights are set to 26.6 mm, which the SolidWorks model claims sets d1 to the nominal 1 mm. PURPLE - 110620 Measurement, with the blade tip heights set to 26.6 mm. One immediately sees that the lowest pitch frequency is lower than expected, at 0.449 Hz. GREEN - Modified model, *decreasing* d1 by 3 mm, to 2mm above the center of mass. This implies a blade tip height of 29.6 mm. We see that this model nails the first pitch mode, though fails to precisely predict the upper two pitch modes. Since the 110620 measurement and model imply that the M1 blade tip heights are too high by 3 mm, we then lowered them by 3 mm, to 23.6 mm, ideally setting d1 back to the nominal 1 mm below the center of mass. YELLOW - 110621 Measurement, with the blade tip heights set to 23.6 mm. Here, we have increased the lowest pitch mode frequency to 0.461 Hz, and better matched the upper two frequencies to the nominal BLUE model. Confused as to why we didn't get all the way up to 0.48 Hz by adjusting d1 (the M1 blade tip heights), we began exploring other model parameters that might be different from nominal. Norna explored changing all of the d's: d0 - the suspension point to M1 connection at M1, d2 - the M1 to M2 connection at M2, d3 - the M2 to M3 connection at M2, and d4 - the M2 to M3 connection at M3. Note that in reality, only d1 may be adjusted "on the fly," to change the remaining d's would require new mechanical parts. However, varying d4 most accurately replicates what has been measured: RED - Modified model, *decreasing* d4 by 1 mm, to aligned with the center of mass (i.e. d4 = 0 mm). This model implies the prism height, with respect to the center of mass is incorrect. Joe took some physical measurements, and compared them against the solid works model and respective drawings, and while doing so discovered that the prisms (D080583) are version "F" when the production units should be at version "G". Although this needs to be fixed, the difference between version "F" and "G" is only in the distance to which it protrudes from the M3 mass, and therefore does not effect d4, and consequently would not effect the pitch mode. Interestingly, if we *increase* the nominal d4 by 1 mm (to 2 mm above the center of mass), the model predicts that we might get a much stiffer first pendulum mode, without effecting the frequency of the upper pitch modes: CYAN - Modified model, *increasing* d4 by 1 mm. This gets us a particularly stiff lowest pitch mode, without compromising the upper pitch mode frequencies. It turns out, that *flipping M3 upside down* (rotating 180 degrees about the transverse axis) will accomplish exactly this increase in d4, according to Joe's calculations using the SolidWorks model. Thus, in order to confirm that d4 effects this particular transfer function -- and to get another data point -- we will flip M3 over tomorrow (a reportedly simple task), and remeasure the pitch mode. We have no intention of making this change to the production units, as (we believe, though it has not yet been confirmed) it's too late to make such a change to the glass BSs and/or FMs, and given that the "requirements" for this particular mode are defined in away that is merely to get the first pitch mode away from the first longitudinal mode to simplify local damping. Stay tuned!! Note that we have strong evidence against two other "problems" (1) The modeled M1 blade spring stiffness is incorrect. We measured a separate M1's 4 blades, and their stiffness matches the model within the uncertainty of the measurement. See sub-entry to follow. (2) The trim mass distribution of M1 is unbalanced. Joe redistributed the trim mass to be more balanced, and a quick transfer function, and subsequent overnight spectra revealed no change in the lowest pitch frequency.
J. Kissel, J. O'Dell, R. Lane, T. Sadecki
In order to rule out a variant in the BSFM modelling, that might explain the unexpected results from X1 SUS BSFM01 pitch transfer functions, we quickly measured the vertical stiffness of 4 M1 blade springs on another assembled M1 mass. The model predicts a vertical blade spring stiffness of 1460 N/m, and the measured stiffness for each blade matches the model to within the measurements' uncertainty. This rules out the possibility that an error in the model's blade stiffness is causing the discrepancy between what has been measured and what is modelled.
Details
----------------
Measurement Setup (see second attachment): with a M1 mass dog clamped to an optical table, we loaded a single blade with the nominal weight of the lower stages experienced by a single blade. Then, we measured the displacement of the blade tip using a dial indicator after adding (and then removing) 50, 100, and 150 grams of trim mass to the load (and took the mean).
The Physics: It's always a good day when you can fall back on good 'ol Hooke's law, F = (-)k . x. We calculated the stiffness using
F = k x
m g = k x
k = m g / x
and calculated the uncertainty in stiffness, assuming a 0.002 in uncertainty in displacement, and that the uncertainty in force (and g) was negligible:
k = F / x
dk^2 = |dk/dF|^2 dF^2 + |dk/dx|^2 dx^2
(dF == 0)
dk^2 = |dk/dx|^2 dx^2
dk = |dk/dx| dx
( |dk/dx| = F / x^2 )
dk = (F / x^2) dx
where the displacement uncertainty came from a sticky dial indicator, and any addition force that it might have added (or prevented).
The ICC Crew today: Mark Layne, Zack Haux, Chris Soike We started the morning by inspecting the compressors with John W. and Michael L. The ICC crew had noticed moisture in the air hoses attached to the compressors. After checking the hoses and the filters, John was convinced that the moisture was water condensate and not a threat to our work. He did suggest blowing out the lines both before the beginning and end of work each day so that will be added to the procedure. John and Michael were in agreement that work should proceed. Tasks accomplished today -West beamtube dust barrier (iLIGO) installed -Gaps at the top of north and south dust barriers were filled in -Condition of chamber documented (pix) -Pre-work particulate samples taken -Particulate depletion samples taken -Soft dome cover removed and soft roof retracted -2 sections of the collar area cleaned -Soft door/dome covers re-installed -Hoses blown out at end of work -Materials consolidated to provide room for HAM 5/6 door lay-down
Leaving GV5 soft-closed
- Cleanroom moving to HAM5, bolts taken off north door by OMC TT1 - Quad 3 being moved to EY - Greg to break cement under HAM9 piers because aLIGO piers are slightly bigger than eLIGO piers - Squeezer electronics being moved to between HAM5 and HAM6 - Gregorio Tellez, Eric Black, and Eric James, inspect HEPI piers in LVEA - h2adcumy needs to be rebooted and it's MEDM needs to be fixed so that it shows red not black. Black is ambiguous. - Reibold, MikeL, MattE to EY for penetration and drilling. Vacuum cleaner brought to keep dust down - Kyle vents vertex and HAM6 - Deliver from Conway for Hugh to the warehouse - Oxarc delivery for Kyle - Small rodent found under Control Room big table during safety meeting. It crawled under the FMCS workstation and disappeared...it will return, no doubt. - SUS people working on test stand in the LVEA - LVEA is currently transitioning to laser hazard. Keita will lock mode cleaner (?) and Squeezers will work from there. - MattW will be leading efforts at the lab in EY. The phone extension there is x175 (posted on white board).
(Corey, Jeff, Jim, Mick V.)
Today consisted of lots of helicoiling and torquing. One team torqued down the Stage1 Close-Out Plate, while another team (mainly Jeff) worked on helicoiling the Top-Facing Keel Plate. Once the Stage1 Close Out Plate was torqued down some other Gussets and hardware were installed on the system (such as the Outer Upper Walls for Stage2.
The Bottom-Facing Keel Plate was installed on the system. Once the Top-facing Keel Plate was helicoiled, the day ended with us dropping this Keel Plate on it's respective plate. We now have all big plates on Assy#2!!
A background activity was building the Sensor Assemblies.
Sub-Assembly Status:
Jonathan and Dave.
We have installed ligo.org authentication on all the H2 CDS Workstations in the control room and the LVEA. In addition to the "controls" generic account, you may log in using your "albert.einstein" ligo.org account. Note if you dont already have a CDS account I will need to create your home directory before this will work for you.
Access to CDS from outside of CDS is still strictly controlled and is not ligo.org authenticated. If you need a remote access account, please contact me.
Please email Jonathan and myself if any problems are found with the new scheme.
(Corey, Eric, Jeff, Jim, Mick V.)
Stage2 Work
After part of Stage1 was put on the Optics Table, continued to build up Stage2. The Mid-Plate was installed, Upper Hex Assembly installed, and then we moved back to Stage1 work....
Stage1 Work
After the Stage2 Mid-Plate/Upper Hex Assembly was installed, the Stage1 Close-out Plate & Close-Out Plate Cover were installed (still need to put in bolts and torque the latter).
Subassembly Status
Misc.
Top-Facing Keel plate was put on the Granite Table (lots of helicoiling for it!)
Gerardo assisted- Dumped gate annulus then opened GV10. Dumped gate annulus then cycled GV18, cycled GV5, dumped all annulus volumes then cycled GV1, cycled IP1, IP2, IP3, IP4, and IP11.
PT144 alarms should now go away
Sprague Unifirst Mid Columbia Forklift delivery from Peninsula BSC7 dome removal for in chamber cleaning LVEA SUS test stand assembly Transition to laser hazard for locking of H1 Y arm Optical lever mid arm testing
BSC-7 The dome was removed by the Apollo crew, led by Randy Thompson. The operation went pretty smoothly although the dome still lists to one side and counter-weight must be used to keep it level during transit. The o-ring protectors were installed on the flange (6 sections)and C-3 soft covers were installed on both the door and the dome. Mark Layne, Zack Haux and I garbed up after lunch. Mark and Zack removed the one electrical feed-through on the chamber and it was wrapped for storage. Zack went into the chamber and installed support tube/bellows protection (C-3 covers)on the west support tube. Mark prepared the 60" dust barriers and Zack installed them in the north and south beamtubes. We thought that the west beamtube nozzle might be a different size than the others but it wasn't true. Turns out that once we got some experience with installation, the dust barrier fit much better. We still have a little work on that dust barrier as the Viton rim is lost somewhere on site. (If nothing else we'll cut and clean a new one.) The other tooling is ready to go tomorrow: the compressor trailer is in the high bay (close to the BSC), the hose is there too, six tools are cleaned and in the staging area-two with shrouds attached so we can get started tomorrow.
BSC-7 20 June 2011 Continued training Apollo contractors (Mark Layne and Zack Haux) regarding contamination control and viewports (Kyle Ryan), general in-chamber cleaning procedure, and job hazard analysis (Michael Landry). Removed four viewports, one camera, and one illuminator. Viewports were preserved in Class A condition. Camera and illuminator were wrapped in UHV foil and set aside.
J. Kissel, R. Lane While working on the BSFM we came to the conclusion that we needed to be able to plot the RMS values of the OSEMS. This included some slight modifications to the model to include new Test Points, recompilation of the model, modifications to the.ini to be able to record the new RMS Test Points, and a DAQ reload. We took a power spectrum of X1:SUS-BSFM_M1_OSEMINF_F1_IN1_DQ and X1:SUS-BSFM_M1_WD_OSEM_F1_RMS_IN1_DQ (CDS RMS Channel), discovering that the CDS RMS filter is, in effect, a Low Pass Filter. Note: X1:SUS-BSFM_M1_OSEMINF_F1_IN1_DQ(RMS) is the DTT RMS of X1:SUS-BSFM_M1_OSEMINF_F1_IN1_DQ.
(Corey, Eric, Mike V., Jim)
Plate Shuffling
Thought we could work on the Stage1 floor on the Assembly Stand, but since we need to install Ballast Weights under it, it wasn't going to work. So this plate was moved onto the Granite Table (where the Stage2 Midplate was....Stage2 Midplate was moved on to the Assembly Stand).
Stage0 Assembly
5/8" bolts were torqued down (once we received a Class-B-ed socket).
Stage1 Assembly
After the Stage1 Floor was moved onto the granite table, the walls were installed. This assembly was then installed on to the top facing Optics Table plate.
Stage2 Assembly
Tried to be clever and pre-assemble the inner hex walls on this plate while it was on the Assembly Stand, but we discovered these walls cover up holes for other bolts we need to get to later, so the inner hex was disassembled.
B. Bland, J. Kissel, R. Lane, J. O'Dell, N. Roberston, T. Sadecki Over the weekend, I took a two vertical transfer functions, at high frequency (15-50 Hz) and at low frequency (0.05-0.3 Hz), using swept sine DTT (as opposed to the white noise excitations that have been performed in the past) just to begin to get a feel for what would be excitation levels will be needed to get good coherence for Matlab transfer functions. However, in the process, I discovered that there was a good deal of cross-coupling between vertical and roll (with a little bit of transverse -- see first attachment). This cross-coupling is completely unexpected, but our best guess was that it had to do with the recently re-adjusted blade spring tip heights -- in that one side of blades, or even a particular blade was higher that the opposing side. So, taking advantage of the newly arrived Joe O'Dell, we went in and re-assessed and re-adjusted all four blade spring tip heights, such that they "better" matched the nomimal 26.6mm from blade spring tip height to M1 base plate (see description of measurement in Saturday's aLOG). After a few more lessons from Joe on how best to align the OSEMs (use CAM tools instead of just fingers -- see second attachment), we sat down for another round of transfer functions. Retook the 15 -50 Hz Vertical transfer function (because it was quick), and the cross-coupling has *changed* but *not improved.* Both Joe and Norna recommended a pitch transfer function to diagnose whether we've got the blade tip heights correct. Attached (fourth attachment) are the results of the (white noise) transfer function, compared with two models. The first, blue, "d1 = Nominal" model uses the same parameter set that was used for the BSFM's final design review (see T1000724), except that I've removed the damping by turning the damping gains to 0.0, to recreate a free model. This is what the M1, Pitch to Pitch transfer function *should* look like. The second, green "d1 = Nominal - 3 mm" model uses the exact same parameter set, except for subracting 3 mm from the d1 parameter (the distance between the vertical suspension break off points of the M1 blade springs and the M1 center of mass). This is the same as *raising* the blade spring tip heights by 3 mm. One can see that the "d1 = Nominal - 3 mm" model matches first two measured resonances exactly -- the first mode at 0.41 Hz is a L mode. The second mode, at 0.44 Hz (nominally at 0.48 Hz) is the pitch mode that we're trying to nail (see M080134 for reasoning). Again, this implies that the blade spring tip heights are *too high* by 3 mm. So, tomorrow we will readjust the blade spring tip heights such that our fiducial measurement reads 23.6 mm (the distance between the blade spring tip and the M1 base plate). The DTT files for the attachments can be found here: ~/SusSVN/sus/trunk/BSFM/X1/BSFM01/BUILD02/SAGM1/Data (First Attachment) 110617_X1SUSBSFM01_SweptSine_V_15to50Hz.xml 110617_X1SUSBSFM01_SweptSine_V_0p05to0p3.xml -- This one's cycles and averages are ridiculous, if you want to use it again (and have it *not* take all weekend, then reduce the number of cycles / averages / points) (Second Attachment) 110620_X1SUSBSFM01_OSEMDiagonalization_VtoLFRT_1p1Hz.xml 110620_X1SUSBSFM01_OSEMDiagonalization_YtoF2F3_1p4Hz.xml (Third Attachment) 110620_X1SUSBSFM01_SweptSine_V_15to50Hz.xml (Fourth Attachment) 110620_X1SUSBSFM_WhiteNoise_P_0p005to50Hz.xml (Exported as 110620_X1SUSBSFM_WhiteNoise_P_0p005to50Hz_F1F2F3LFRTSDLTVRPY_{tf,coh}.txt) The Matlab script to process the Pitch transfer function is plot_110620_X1SUSBSFM01_WhiteNoise_P_0p005to50Hz.m (which uses the BSFM model and parameter files in ~/trunk/BSFM/Common/MatlabTools/BSFM_Model_Production/, which are edited versions (as described above) of T1000724)
(Corey, Eric, Fabrice, Greg, Jeff, Jim, Mitch, Myron, Sebastian, Vincent, et. al.)
It's been a while since progress has been posted, but tons of work has occurred; so here's a rough summary of the main assembly points of the last week or two.
BSC ISI#1
Notes: The L4C's & GS13's we recently received from LLO had noticeable dark powder/marks on them (believe this is the typical Aluminum Oxide we've seen before).
WHAT'S NEXT ON #1: Goal is to float the assembly today (which means tossing lots of mass on the Assembly!), and hopefully start some testing soon (Vincent could atleast play with the Horizontal GS13's)
Circuit Overload?
We have been overloading one of the electrical circuits in the Staging Building (noticed this when we hooked up Rai's helicoil machine and vacuum). When this circuit trips, it would shut every thing on this circuit down: including both the BSC/HAM Test Stand AND SUS Test Stand (!!!). Several times Vincent had to deal with losing all of his work and it would generally take a few hours to get everything back on-line.
I believe there are the two plugs which are on the wall with the circuit in question. Do not plug anything into these circuits which could overload the system (i.e. vacuum for example). Perhaps we should put up some signs, or tape over the plugs, or have some killer moths guard the plugs (more on them below).
BSC ISI#2
Roughly 75% done with helicoiling Optics Table
Staging parts for this assembly
Attack Of The Moths (& Tropical Temperatures)
Over the last month or so, we've noticed a deluge of moths within the cleanroom area of the Staging Building (attached is a photo of a pile of dead ones on the floor). They've been spotted inside and outside of our big cleanrooms (and we found a dead one on a large BSC plate under a cleanroom cloth cover!). This is just to note if we come across any questionable particulates.
On Monday it was downright tropical in the Staging Building (can't remember but it was in the mid to high 80s). John was able to give us tolerable temperatures by the afternoon (perhaps the moths like warm weather & learned how to futz with our thermostat!)
Shims Used For Locker Installation
(this is from email from Sebastian)
. Shim thickness for the lockers between stage0-1 (big lockers) :
Corner 1 : 0.127"
Corner 2 : 0.125"
Corner 3 : 0.127"
. Shim thickness for the lockers between stage1-2 (small lockers) :
Corner 1 : 0.126"
Corner 2 : 0.122"
Corner 3 : 0.130"