A. Effler, J. Kissel, B. Shapiro

I've reconstructed the filters and gains that have been used to damp the H2/H1SUSETMY for the H2OAT, calibrating them with the appropriate signal chain gains in an attempt to finally merge models with measured reality. 

I attach two sets of plots.
(1) dampingfilters_H1SUSETMY_20121121.pdf: shows the filters, both in their raw uncalibrated (normalized) form (pg 1), as well as in their calibrated form (pg 2) using the EPICs gains that are currently in place.
(2) 2012-11-21_QUADDampedModel_Top2Top_TFs.pdf: shows a comparison between undamped and damped, model and measurement. The measurements used were taken from H1 SUS ETMY's Phase 3b results.

There is now excellent agreement with model and measurement for all degrees of freedom, most importantly with entirely understood frequency responses and gains. The remaining discrepancies arise from the differences in the undamped modeled plant and measured plant, namely the high frequency zero we believe to arise from drive and sensing chain cross coupling.

Now that model and measurement agree to such precision, we can now
- Explore compensation for the high-frequency zero
- See that this current set of filters and gains over damps most degrees of freedom, most likely reinjecting too much sensor noise into the system for a fully operational aLIGO IFO
- Accurately predict sensor and actuator noise at the test mass and design filters that will meet aLIGO requirements accordingly 
- Comfortably design hierarchical control filters which take into account realistic damping loops.

--------------------------------
Details:

Why we hadn't gotten here before today:
There were three pieces missing from the puzzle that had previously resulted in inaccurate models. 
(1) Prior to this update, the model had still been using legacy filters (in frequency response alone), and was using gains fudged by hand to get roughly the desired damping. This was "good enough" for the purposes needed to date, so we hadn't put much brain power into it. Perhaps more accurately I'd been previously stumped by the bug in the connection matrix described here, which is now fixed.
(2) The filter frequency response, along with using the actually-installed-EPICs gains, were were designed in foton "by hand." I've now exported and reproduced them in Matlab. The filters are calibrated into [N/m] or [N.m/rad] using the signal chain described briefly below (and in detail in LHO aLOG 4756 and LHO aLOG 4563). This calibration is needed because the model is (and has always been) calibrated in physical, SI units. The model of the closed loop is therefore

                                                             +--[m/N]--+
Ext. Forces -[N]-->(+)---------------[N]---------------------|  Plant  | ----------------[m]---------------------+-----> 
                    |                                        +[rad/N.m]+                                         |
                   [N]                                                                                          [m]
                    |   +-------------+              +----------+  +---------+                 +-------------+   |
                    +---|    Drive    |<--[drv. ct]--|EPICS GAIN|--|  DAMP   |<-(+)-[sns. ct]--|    Sense    |<--+
                        +-[N/drv. ct]-+              +-[ct/ct]--+  +-[ct/ct]-+   ^             +-[sns. ct/m]-+
                                                                                 |
                                                                                 +-- EXC

where the DAMP filter units, [ct/ct], are the designed, normalized frequency response. Note, 
- This reconstruction was done using the function
${SusSVN}/sus/trunk/QUAD/Common/FilterDesign/reconstructquadfotonfilters_20121121.m, where you can see the details of each filter design. I've saved the filters in the file
${SusSVN}/sus/trunk/QUAD/Common/FilterDesign/dampingfilters_QUAD_20121121.mat

- In practice, thanks to the powers of linear algebra, instead of treating the sensor chain (Sense) and actuator chain (Drive) independently, I merely multiply the normalized filters by the total DC gain of the signal chain,
 
              % <----------- Sensor Chain [sns. ct/m]------------>   <--- Actuator Chain [N/drv. ct] --->;
              %     OSEM                     SatAmp                              Coil Driver Coil/Magnet ;
              %  Sensitivity                TransImp.    ADC            DAC       TransCond. Force Coeff.;
              %(   [mm/uA]    [m/mm]  [uA/A]   [A/V]   [V/sens ct]   [drive ct/V]    [V/A]      [A/N]     )^-1= [(sns. ct/m) . (N/drv. ct)];
calibration = (( (0.7/76.29)*(1/1e3)*(1e6/1)*(1/240e3)*(40/2^16) ) * ( (2^18/20)*(1/0.009943)*(1/1.694) ))^-1;
            = 55.071 % [(sens ct/m) . (N/drive ct)] or [(sens ct/rad) . (N.m/drive ct)] 


(3) Previously, the damping filter's overall gain had been positive, and the feedback negative sign had been taken care of in the connection matrix, where the filters were hooked into the undamped state space model, in
${SusSVN}/sus/trunk/QUAD/Common/MatlabTools/QuadModel_Production/generate_QUAD_Model_Production.m.
Now that I'm using the actual EPICs gains, where the negative sign is explicitly called out, the sign in the connection matrix has been removed. I mention this only because it had stumped my for a day (let it be a less kids: always pay attention to the phase of your transfer functions!). Note this means all future filters input to generate_QUAD_Model_Production.m must have the overall sign be negative (regardless of whether it's included in the filter itself or explicitly in the EPICs gain).

These changes have been committed to the repository, with these 20121121 filters currently set as default; please svn up!

Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot from approximately 6 PM Nov. 25 to 6 PM Nov. 26. Also attached are plots of the modes to show when they were running/acquiring data.

The sensor may be failing on the dust monitor at location 2 in end Y. The local status was recorded in alarm a number of times today. It is not currently alarmed.

CP7 dewar's vapor pressure value not responding to regulator adjustments -> fully closing (CW) adjustment screw has no effect -> fully opening (CCW) adjustment screw has no effect -> exhaust flow is continuous and not adjustable -> Closed manual valve which is in series with regulator to stop exhaust flow -> used heat gun to thaw exterior of regulator -> opened manual valve and fully closed and fully opened adjustment screw -> no change -> replaced regulator and left adjustment at 0 psi < factory setting < 30 psi -> manually filled CP7 to ~100% using LLCV bypass valve (resulted in CP7 alarm) 

Conclusion: Discovered upon disassembly of removed regulator that it was full of water.  This would have been ice with GN2 flow and explains why the adjustment was prevented as my thawing of the exterior wasn't enough to thaw the regulator's interior.  Otherwise the regulator was normal-no broken or worn-out parts.  The two threaded halves of the regulator assembly don't have and effective water seal between them.  When installed with the adjustment screw pointing up, as this one was, liquid water which condenses on the surface of the regulator during the summer months can enter the interior of the regulator and collect.  I will re-orient any regulators found to have their adjustment screws pointing up such that they point down.  

Expect CP7 alarms over the next day or two as the system comes into a new equilibrium and minor tweaks are made.

HEPI pump stations #5 and #6 have gone through the initial start up phase this afternoon and are now pumping 419-TY.  The pump servo set point is at 30psi to recirculate locally on the mezzanine driving pump stations #5, #6, #7, #8 at 20.5 HZ=615 rpm.  Pump stations #7 and #8 were started 1yr 3mo to 1yr 4mo ago.  

WP#3576

The IOP WD bypass time was extended from 10 mins to 6 hours for h1sush34 (MC2, PR2, SR2). This will permit mode cleaner commissioning which requires large DC offsets to be applied to MC2. This WP will remain open until the bypass time is restored to 10 mins.

GregG, JimW

After software issues were sorted last week, we were able to finish roughly balancing the ISI this morning. CPSes still need to be re-gapped and the balancing should be fine-tuned, but as of now the locked position is where it needs to be, after HEPI adjustments from last week, and the unlocked position is only a couple milli-inches off from that. The ISI was left locked, so SUS is clear to start their work.

Cycled instrument air to LLCV and verified that needle was 100% open as indicated by CDS -> Temporarily valved-in magnahelic differential gauge and confirmed CP7's level, 38" = 92% full -> Noted indicated vapor "head" pressure for LN2 dewar was low, ~5 psi -> No indication of parallel paths for vapor exhaust -> Adjusted exhaust pressure regulator 1/2 turn CW.

LLCV nominally 75% open to maintian CP7 level at setpoint -> Presently maxed-out at 100% -> Am monitoring indicate pump level -> may or may not investigate tonight 

Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot from approximately 6 PM Nov. 20 to 6 PM Nov. 21. Also attached are plots of the modes to show when they were running/acquiring data.

Pump controller should have come on-scale by now if gas load was due to O-ring flanges (now isolated from pump) -> Swapped controller -> leaving pump valved-out -> leaving aux. cart pumping annulus (1.3 x 10-6 torr)

I've taken long-overdue damped and undamped, Phase 3b (with BSC6 at vacuum), transfer functions H1 SUS ITMY's main (M0) chain (formerly H2 SUS ETMY). Attached are the results: 2012-11-19 data is damped, and 2012-11-20 undamped. No surprises -- all mechanics look great / as expected. In processing these measurements, I've made improvements to the QUAD analysis scripts (see details below), so please svn up the QUAD corner of your local copies of the SusSVN. These were taken in-between Vincent's commissioning of BSC6-ISI, so I didn't press my luck to try and get reaction R0 chain measurements. We'll get 'em eventually.


Notes: 
- The damped transfer functions use the "final" set of filters and gains that were tuned / used during the H2 OAT, which look to be a little bit too aggressive, but it's a totally "by feel," qualitative assessment. With the newly fixed damped QUAD model, I intend to slap these filters and gains in (instead of the old "legacy" filters and arbitrary from LASTI which are in there now), and predict the test mass displacement -- specifically how much arises from sensor noise re-injected into the SUS vs. residual seismic noise. Stay tuned!

- The data was taken with freshly minted H1 SUS ETMY DTT templates, which can be found here:

/ligo/svncommon/SusSVN/sus/trunk/QUAD/H1/ETMY/SAGM0/Data/
2012-11-19_1840_H1SUSETMY_M0_Mono_L_WhiteNoise.xml
2012-11-19_1840_H1SUSETMY_M0_Mono_P_WhiteNoise.xml
2012-11-19_1840_H1SUSETMY_M0_Mono_R_WhiteNoise.xml
2012-11-19_1840_H1SUSETMY_M0_Mono_T_WhiteNoise.xml
2012-11-19_1840_H1SUSETMY_M0_Mono_V_WhiteNoise.xml
2012-11-19_1840_H1SUSETMY_M0_Mono_Y_WhiteNoise.xml

(The same template is used for both damped and undamped transfer functions, so the date is the only thing one should need to change.)

- I've made several improvements to 

${SusSVN}/sus/trunk/QUAD/Common/MatlabTools/
plotquad_dtttfs.m
plotquad_matlabtfs.m
plotallquad_dtttfs.m
 
(1) I've added a few lines at the end of all three scripts which removes the individual plot .pdfs after they've been merged into the ALL.pdfs. They're just clogging up the svn, and were never use them anyways. 
For the "raw measurement" processing scripts, I've
(2) In the interest of getting all the gains right for predicting the damping loop performance, and because we now know the calibrations of the sensor and actuator chains well from independent measurements, I've changed the way the scripts calibrate the data. Namely, I explicitly calculate the [(sensor ct/drive ct) / (m/N)] (or [(sensor ct/drive ct) / (m/N.m)], [(sensor ct/drive ct) / (rad/N)], [(sensor ct/drive ct) / (rad/N.m)] as need be), instead of using the empirically measured value of 60. The number turns out to be 55.0708 [(sensor ct/drive ct) / (rad/N.m)], only an 8% difference, but it also aids in the understanding of improvement (3).
(3) I've added a switch inside the calibration step which -- based on a new, user-input, boolean, variable meas.sensCalib -- determines whether the OSEMINF "to_um" filter has been left ON during the measurement or not. Here's the new lines:

        if meas.sensCalib
                          % <-Sensor->   <---------- Actuator Chain ---------->;
                          %    OSEM                    Coil Driver Coil/Magnet ;
                          % Sensitivity       DAC       TransCond. Force Coeff.; 
                          %   [m/um]      [drive ct/V]    [V/A]      [A/N]   = [(m/um) . (drive ct/N)];
            calibration = (( (1/1e6) ) * ( (2^18/20)*(1/0.009943)*(1/1.694) ))^-1;
        else
                          % <----------------- Sensor Chain ---------------->    <--------- Actuator Chain --------->; 
                          %     OSEM                     SatAmp                              Coil Driver Coil/Magnet ;
                          %  Sensitivity                TransImp.    ADC            DAC       TransCond. Force Coeff.; 
                          %   [mm/uA]    [m/mm]  [uA/A]   [A/V]   [V/sens ct]   [drive ct/V]    [V/A]      [A/N]   = [(m/sens ct) . (drive ct/N)];
            calibration = (( (0.7/76.29)*(1/1e3)*(1e6/1)*(1/240e3)*(40/2^16) ) * ( (2^18/20)*(1/0.009943)*(1/1.694) ))^-1;
        end

Where, for the sensor chain, I've used the calibration as (best) described in LLO aLOG 4291, and for the drive chain, I've used the calibration as (best) described in LHO aLOG 4563 (though the force coefficient is different because HSTSs use 10Dx5T [mm] magnets, and QUADs, BSFMs, and HLTSs use 10Dx10T [mm] magnets on their top stage; the coil driver trans conductance is different because QUADs use QUAD TOP drivers, where HSTSs use Triple TOP drivers). Or, if you like pictures / diagrams, see T1100378. 

In case you're worried, we don't have to go back and re-run all of the old data. It's only an 8% change in the calibration, which we know can be less than the mechanical/electronics gain variation between suspensions. I will say though, I re-ran a few older measurements just to test it out, and it does put the measurements right smack on the model now for some of them (nice!).

GregG, MitchR, JimW

Following the initial float of HEPI and Doug's follow up optical survey yesterday, HEPI was adjusted to fix the level of the the optical table this morning. The adjustment was relatively problem free, and the ISI is now level to within .1mm and is within .1 mm of the required elevation, at 244.1mm on Doug's ruler (see post 4752). After HEPI was adjusted, I locked the feet down while Greg and Mitch dropped cables down and plugged CPSes in. Unfortunately, a cable on a vertical sensor broke during install, so we will have to replace it, before proceeding with floating the ISI. We should be able to finish up by COB today, allowing SUSsers to resume work on Monday.

Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot from approximately 6 PM Nov. 19 to 6 PM Nov. 20. Also attached are plots of the modes to show when they were running/acquiring data.

The plots of counts for the LVEA are missing. There were errors trying to plot them. They were not running for a time yesterday, (the power was unplugged, they probably glitched, and I didn't check for a while to see if they had come back). When they were being rebooted the channels for the calibrated counts were '-nan'. The edcu probably can not record this.

Attached are plots of dust counts > .3 microns and > .5 microns in particles per cubic foot from approximately 6 PM Nov. 18 to 6 PM Nov. 19. Also attached are plots of the modes to show when they were running/acquiring data.

- Keita, Kiwamu, Cheryl


Mode Cleaner:
The beam is centered as it comes through MC1 and centered on MC2.  We were able to center the beam on MC3 using the alignment sliders on MC2.  The beam made it back around to MC1 and is separated from the input beam by about 1cm.  Then it was time to go home.

The alignment was made using irises placed on HAM2 and HAM3 according to the dimensions of alignment tools that Luke made.

One thing we ran into was that one OSEM on MC2 was basically railed, so the watchdog kept tripping.  When MC2 tripped, it moved mostly in yaw which suggests it might be bumping into an EQ stop.


Other work in HAM2:
The beam from the PSL had to be adjusted from inside the PSL, because the centering on the upper periscope mirror was off.  After centering on the upper periscope mirror, the periscope had to be rotated to align the beam to the lower periscope mirror.  The rotation was maybe 1-2 degrees.  The position and angle of AROM RH1 and ROM RH1 were adjusted.  The clamp on ROM RH1 was moved 180 degrees, after seeing that the screw in the clamp was at the far end of the slot, and after finding available space on the top level HAM2 layout drawing.

(Margot P, Gerardo M)

Both primary prisms were finally removed from PR3-01, the adventure started on Wednesday, 11/14/2012.  After roughly 32 hours of soaking (in acetone, a very rare commodity around LHO lately), both primary prisms are off.

The only secondary prism was removed on Thursday, within 9 hours of soaking.

PR3-01 is now ready for its prisms to be glued.

Transferred wall target 501 elevation to new location for better line of sight for the autolevel. I named it 501B
I mounted a scale vertical next to 501B (501B = -26.9mm below Z0 at BSC2, beam tube centerline) and centered the scales 300mm line with the new wall target horizontal line.
I set the autolevel zero on the vertical scale at 500mm (200mm above 501B)
Panning over to BSC2 to the elevation height gage, the scale mounted to the precision rod which Jim W. held against the ISI optical table surface, the autolevel zero read 244.0mm near to the south side of the table.
Note: the distance down from the ISI table requirement is 1661.7mm to the beam tube centerline. So height gage length = 1488.6mm [1244.6mm (49.0") + 244mm on scale = 1488.6mm] + 200 on wall target + -26.9 = 1661.7mm 

Results:

HEIGHT: AVERAGE FROM SCALE 243.45mm (244mm = PERFECT ELEVATION)REQUIREMENT +/- 1.0mm FOR BEAM CENTERLINE so the closer the better (<1.0mm ?)

LEVEL REQUIREMENTS: ACCURACY REQUIREMENT:+/- 100urads (0.1mm DIFFERENTIAL HEIGHT) 

SCALE READINGS:
SE CORNER: 243.0 RAISE UP 1.0mm 
SW CORNER: 243.0 RAISE UP 1.0mm
NE CORNER: 243.6 RAISE UP 0.4mm
NW CORNER: 244.2 LOWER DN 0.2mm

WP#3574

Greg, Dan, Dave.

LDAS upgraded the QFS writer for the H2DAQ primary system (h2ldasgw0 machine) to Solaris11.0 this afternoon. I have just remounted this file system on h2fw0 and h2nds0. The h2fw0 is writing frames to the QFS file system.

A bad 2GB mem DIMM was found in h2ldasgw0, this was replaced with a spare. Also, one of the fibre channel cards is showing a fault and we are running on the second card.

WP#3571

The SUS and ISI models to support BSC1 install work were started today. These were non-standard installs due to the following constraints:

• all Dolphin IPC was removed from the new models, we have hit the 64 channel limit imposed by RCG2.5.1
• no new channels were added to H1 DAQ, we have maxed out the DC and NDS memory

Both of these limitations are temporary. When we upgrade to RCG2.6 the IPC limit will be raised. Tomorrow I will receive more memory for the H1DAQ (most probably wont install till next Tuesday).

I created IOP models for the SUS and SEI BSC1 frontends (h1susb123 and h1seib1). These were copies of the H2 equivalents minus the Dolphin IPC.

I copied H1SUSETMY model to become H1SUSITMY. The appropriate DCUID/computer changes were made and all Dolphin IPC was removed.

I copied the H1ISIETMY model to become H1ISIITMY. Again DCUID was changed and Dolphin IPC was removed. We had some confusion about the ADC distribution between HEPI and ISI. The models we have been using give ISI the first three ADCs and HEPI the fourth. Drawings indicate the opposite, HEPI getting the first ADC and ISI the last three. We will work on resolving this with team SEI.

As stated earlier, no H1DAQ changes were made. None of the BSC1 channels are being recorded by the DAQ.

Following a similar prescription as what Keiko's done at LLO (see e.g. LLO aLOG 4030), but now using the new infrastructure, I've installed a few gains,

H1:SUS-PR2_M1_OPTICALIGN_P_GAIN 2.636 [drive cts/urad]
H1:SUS-PR2_M1_OPTICALIGN_Y_GAIN 1.859 [drive cts/urad]

into the new OPTICALIGN bank, which should now be responsible for aligning the optic. With this gain, the offsets

H1:SUS-PR2_M1_OPTICALIGN_P_OFFSET (for Pitch)
H1:SUS-PR2_M1_OPTICALIGN_Y_OFFSET (for Yaw)

are then calibrated into [urad] of optic motion. This gain should be [applicable to / the same for] every HSTS.

I attach some shots of the new screens demonstrating where these gains live (and showing off the new screens).

-----------------------------
How the gain was calculated:
It's along signal chain, but here goes (I use Yaw on an HSTS as an example):
                                            
                                                                 LF  +--------+  +------+  +-----------+  +-----------+  +---------+  
                                                                  +->|COILOUTF|->| DAC  |->|Coil Driver|->|Coil/Magnet|->|Lever Arm|->+
+-----------------+  +---------------+  +----------+  +--------+  |  +[cts/ct]+  +[V/ct]+  +---[A/V]---+  +---[N/A]---+  +---[m]---+  |  +-------------+   +------------------+           
|OPTICALIGN OFFSET|->|OPTICALIGN GAIN|->|DRIVEALIGN|->|EUL2OSEM|->+                                                                   +->|HSTS M1 to M3|-->|Optic Displacement|
+-----[urad]------+  +---[cts/urad]--+  +-[cts/ct]-+  +[cts/ct]+  |  +--------+  +------+  +-----------+  +-----------+  +---------+  |  +--[rad/N.m]--+   +------[urad]------+
                                                                  +->|COILOUTF|->| DAC  |->|Coil Driver|->|Coil/Magnet|->|Lever Arm|->+
                                                                 RT  +[cts/ct]+  +[V/ct]+  +---[A/V]---+  +---[N/A]---+  +---[m]---+

(see fullsignalchain.png if you browser sucks at ASCII art)
Thankfully, because 
- we're only looking for the DC gain of this path, 
- the DRIVEALIGN matrix is an identity at this point
- the EUL2OSEM matrix accounts for (divides out) the number of actuators (two in this case) and the lever arm between the OSEM and the center of rotation
- the COILOUTFs are unity at DC
then our calculation only involves this simplified chain:

+-----------------+  +---------------+  +------+  +-----------+  +-----------+  +-------------+   +------------------+
|OPTICALIGN OFFSET|->|OPTICALIGN GAIN|->| DAC  |--|Coil Driver|--|Coil/Magnet|->|HSTS M1 to M3|-->|Optic Displacement|
+-----[urad]------+  +---[cts/urad]--+  +[V/ct]+  +---[A/V]---+  +-["N.m"/A]-+  +--[rad/N.m]--+   +------[urad]------+

(see reducedsignalchain.png if you browser sucks at ASCII art)
which means the OPTICALIGN GAIN is:

OPTICALIGN GAIN [cts/urad] = ( DAC [V/ct] * Coil Driver [A/V] * Coil/Magnet [N/A] * HSTS M1 to M3 [rad/N.m] * 1e6 [urad/rad] )^-1
                           = ( (20/2^18)  * 0.011919          * 0.963             * 0.4332 (for Yaw)        * 1e6            )^-1
                           = 1.850 [cts/urad]

The only difference for pitch is that HSTS M1 to M3 [rad/N.m] = 0.6172 (for Pitch). The electronics gains of the signal chain were taken from T1000061 (the contents of which have been experimentally confirmed elsewhere), and the HSTS M1 to M3 coefficients were taken from the production model (e.g. see T1200404).