Reports until 18:48, Monday 05 January 2015
H1 SUS (FMP, SYS, VE)
jeffrey.kissel@LIGO.ORG - posted 18:48, Monday 05 January 2015 - last comment - 12:31, Tuesday 06 January 2015(15878)
H1 SUS ETMX Main Chain is Rubbing after Pumpdown
J. Kissel, B. Shapiro

The summary says most of it -- we've confirmed with two degrees of freedom of top to top transfer functions. Our best candidate is that the temperature in the VEA is too high. I tried adding a vertical offset in either direction, in hopes that we have enough range to recover the drooping, but it appears we do not. We'll first try restoring the XVEA temperature (if not surpassing it), but we may have to vent again. Cross your fingers.

Details:
Hoping that we could test the newly turned on H1 SUS ETMX ESD for functionailty, Brett and I noticed the optical lever did not appear centered in either the misaligned or aligned state. We could only restore the optical lever centering by putting the alignment offsets at 
                       P        Y       
Force Realignment    +21.3    -122.9
Original Aligned    +417.0      77.2
Change              +395.7     200.1
Further, I noticed that a P request would cause both P and Y motion, and vice versa. 

Betsy then trended the temperature in the VEA, see LHO aLOG 15877, and found it ~2 [deg F] or ~1 [deg C], which corresponds to about 100 [um] sag at the TOP mass (see T1400749, specifically LLO aLOG 15636. I note that this is at the TOP mass, because the lower stages will sag MORE, since there are cascading vertical blade springs.

We then, in the interest of time, took the transfer functions we know are the most sensitive to rubbing: P to P and V to V. These transfer functions are attached, for the various vertical offsets applied; see 2015-01-05_2358_H1SUSETMX_M0_Mono_WhiteNoise_*_0p01to50Hz.pdf. The vertical offsets were +/- 200 000 [ct], equivalent to most of the DAC range, which is roughly +/-115 [um_pk]. We can clearly see that the first several modes have shifted significantly, and several DOFs are cross-coupled in. Notable, however, is the highest-frequency modes are unaffected. This implies that the top-mass is free, and the lower masses are restricted, as seen in a QUAD's mode shapes. This makes sense, because for this most recent cleaning (see LHO aLOG 15744), the *only* activity in chamber was to clamp the test mass briefly for cleaning. Further, sadly, even with 100+ [um] of displacement in vertical, we could not move the the suspension free.

Note, we checked the reaction chain with P and V TFs, and it appears free and clear (see 2015-01-06_0152_H1SUSETMX_R0_WhiteNoise.pdf). We did not check the TMS, since it was not touched.

Finally, because we were amazed that the SUS had sagged more that 100 [um], and that we know that Betsy set the EQ stops when the VEA was at ~70 [deg C], Brett compared the top-mass displacement of the main chain, reaction chain, and TMS to gauge the amount of displacement compared to the other SUS in the chamber, which should have roughly comparable sag because they've the same blade springs and overall suspended mass (roughly). We attach two trends, EX_SAG_21DecTo5Jan.png (a 15 day trend that includes the pump down) and EX_SAG_22DecTo5Jan.jpg (a 14 day trend to zoom in on the long term temperature equilibration). From the 21st, one can see that the removal of air [the first big sharp drop], caused all SUS to drop. However, the main chain is expected to drop 170 [um] (see T1100616), and the reaction is expected to drop 100 [um]. While the reaction chain drops as expected, the main chain only drops ~125 [um], indicating a sort of bottoming out. Further, from the 22nd's trend, we see the temperature dependence is different (the bias has been removed for clarity). 

So, again. Bad news. Hopefully we can pull this SUS back up with temperature!
Images attached to this report
Non-image files attached to this report
Comments related to this report
betsy.weaver@LIGO.ORG - 10:34, Tuesday 06 January 2015 (15887)

Regarding the expected buoyancy shift during pumpdown, the test masses do not seem to sag as much as expected during the pump down.  Attached are plots of previous ETMx, ETMy, and ITMx pumpdowns (circa 2013 and 2014), showing this.

 

In summary, according to the T1100616 buoyancy calculation sited above, the test masses should sag by ~170 um, however the plots show sags of ~120 +/- ~10 um based on where you think the suspension starts and settles to.   Note, the ETMx data from Dec 2013 looks a bit suspicious and I may have chosen incorrect baselines for where the suspension was sitting in vertical height before and after pump down.  It is difficult to decouple the various pumping operations and temperature effects from these plots.

 

QUAD Main Chain Vertical shift data taken from the plots:

ITMx  115--5= 120um

ITMy  180-65 = 115um

ETMx  150-65 = 85

ETMy  215-95 = 120

Images attached to this comment
brett.shapiro@LIGO.ORG - 12:31, Tuesday 06 January 2015 (15892)

To take a closer look at the buoyancy effect during pumpdown, I removed the effect of temperature by subtracting the reaction chain vertical height from the main chain in the ETMX data Betsy posted above. The two chains should respond nearly the same way to temperature. However, they will respond differently to air pressure since they have different buoyancies (lower stages are different materials, e.g. glass vs steel at PUM).

T1100616 says the main chain should sag by 170 microns while an ERM top mass should sag 100 microns wh8ile pumping down. So if we subtract the reaction chain vertical height from the main chain vertical height during a pumpdown we should see a drop of 70 microns. The attached figure shows that the relative sag in red was only 56 microns, 80% of the prediction. The temperature effects look well suppressed since during times of constant pressure (shown in 2nd figure) the differential hieght remains constant while the individual chains are drifting significantly.

So, clearly the predicted buoyancies are not dead on. If we assume the 80% correction on the differetial sag between chains is valid for each individual chain (which may not be true), then the expected top mass sags will be

main chain: 0.8*170 = 136 microns

ERM reaction chain: 0.8*100 = 80 mircrons

CP reaction chain: 0.8*90 = 72 microns

 

This brings the main chain prediction much closer to Betsy's measurements above, though it is still a bit higher.

 

The script that generates the MATLAB figure is

.../SusSVN/sus/trunk/QUAD/H1/ETMX/Common/Scripts/Buoyancy_data.m

Images attached to this comment