J. Kissel, A. Pele, B. Weaver, T. Sadecki

The Message: We believe that something has gone wrong with the F3 and F3 OSEMs on the reaction chain of H1SUSITMX -- the magnitude of the L2L and Y2Y (i.e. the DOFs that are controled by F2 and F3) is a factor of 1.5 or 2.1 too low. All other DOFs are healthy -- T2T, V2V, R2R, and P2P do not show this discrepancy. This should halt closeout until we figure this out.

This L2L and L2Y discrepancy is supported by looking at the F2 and F3 response to L, P, and Y excitation. After respecting the phase, one also finds the overall sign of the transfer function suspiciously wrong (should be 0 at DC, and it's +/- 180 [deg]).
We do NOT believe:
- it is rubbing. All degrees of freedom's resonances are at the correct frequency, matching model and previous data. All DOFs resonances are also well-defined and highly Q'd.
- it is something wrong with the stiffness / mechanical parameters of suspension. Stiffness would only alter the low-frequency magnitude, and mechanical parameters that affect L and Y (a) would also affect resonance frequencies, but more importantly (b) are only dependent on simple parameters such as mass and wire lengths, of which we are using all the same parts as before but for the UIM mass itself (which again, if THAT were the smoking gun, other DOFs and al resonances would be affected).
- it is the calibration. One calibration is used for all degrees of freedom, we would see the same discrepancy in all DOFs if this were the case.
- it is any element of the electronics chain prior to the OSEMs themselves. Thankfully, the two OSEMs under suspect -- F2 and F3 -- control the common and differential mode of related DOFs, so their cables can be swapped. Doing so did not affect the L to F3F3 results, but they did affect Y to F2F3 results. After swapping the cables, The Y2Y transfer function got *worse* and matched the L2L discrepancy.
- it is an error of the digital parameters. Gains and signs of OSEMINF, OSEM2EUL, TEST, EUL2OSEM, and COILOUTF filter banks and matrices have all been double checked today.

Attached are two sets of comparisons. 
The first shows all DOFs compared against previous instances of this QUAD, and a comparison with an LLO equivalent. 2013−11−22 is a healthy L1 chain. 2013-11-27 is H1SUSETMX healthy, prior to the ITM swap. 2014-06-27_1530 is today's measurements with F2 and F3 correctly connected to the digital system. 2014-06-27_2117 is today's measurement with F2 and F3 cables swapped at the OSEM in vacuum.
The second is 3 pages comparing the L to F2F3 TFs, and 3 pages comparing the Y to F2F3 TFs, in the same order. The first of the three shows the expected results from the 2013-11-26 data set, the second shows the initial discrepant results from 2014-06-27_1530. The third, from 2014-06-27_2117. 

Arnaud has launched overnight matlab transfer functions, in case there're more clues buried in a cleaner measurement. As a reminder, the main chain looks completely healthy, and everything has been IAS blessed, so this is (sadly) the last mystery before close-out begins.

Ideas to try on Monday:
- Redo OSEM open light current assessment, to see if the LEDs have somehow lost half their power
- Swap out F2 and F3 OSEMs for fresh OSEMs.

New data can be found here
/ligo/svncommon/SusSVN/sus/trunk/QUAD/H1/ITMX/SAGR0/Data/2014-06-27*.xml
and exported to similar names, and captured in the .mat file
/ligo/svncommon/SusSVN/sus/trunk/QUAD/H1/ITMX/SAGR0/Results/2014-06-27_*.mat

New versions of
/ligo/svncommon/SusSVN/sus/trunk/QUAD/Common/MatlabTools/
plotquad_dtttfs.m
plotallquad_dtttfs.m
have been committed to the SusSVN. The former has had improvements made to the plotting routine, and the latter has just had these new measurements added to the measurement list.