For the past few months we have been suffering from an unwanted stage 1 Z to stage 1 rZ coupling (the attached plot Z to Z Subtraction shows an example). It seems to have two component which show up in the T240 rZ signal.
There is a 1/f^2 component (in Force to displacement units, the plot is in force to velocity) and a flat component (again in Force to displacement)
The 1/f^2 part does not seem to represent center of mass motion of the platform, as shown by the optical levers (the black line in the Z to OpLev Yaw plot), it is not clear
for the flat component yet.
Speculation for the 1/f^2 has centered on magnetic coupling or local deformation of stage 1, current measurements and FEA on these two effects are off by about the same amount.
Since we believe that the signal is not CoM motion we are going to subtract it from the T240 rZ signal. The plot Z to Z Subtraction shows the predicted amount of subtraction based on the fitted filter.
where the red line is (TF - FIlter)/TF so we are hoping to get a factor of~ 50 subtraction.
The plot Z to OpLev Yaw shows some results.
The BLUE lines shows the transfer function from ISI stage 1 Z drive to the optical lever (solid line is yaw, dotted it pitch) while the ISI is in high blend with no T240 in the control path. So we expect to see
any mechanical coupling.
The PURPLE lines shows the same transfer function with the stage 1 of the ISI in low blend (~40mHz) and the T240s in the loop. Whatever extra pickup exists in the rZ T240 signal will be imposed onto the motion of the ISI
and show up in the optical lever signal (pitch as well as yaw because the suspension is not at the center of the ISI table and there is a translation to pitch coupling).
The Black lines are the same transfer functions with the subtraction path enabled (i'm not even going to try and explain how we got 3 orders of magnitude of suppression)
One disclaimer I calculated a filter gain of 3 and ended up using a gain of 3.24
This is the subtraction results from ETMY
Also I've changed the basis matrices back to what they are supposed to be a defined in T1000388, I used the script Populate_ETMY (X)_Matricxs.m which is in the MISC folder under Scripts
It is just a reduced version of the Populate_MEDM_Screens function
The OPLEV Yaw signal shows the expect factof of ~50 improvement
Some Performance Data on ETMX and EtMY (both are running level 3 controllers with TBetter Blends)
the only new thing here is that the optical lever signals are mostly in the noise. Since we are looking at in loop sensors the injected noise would have been suppressed, so what is new is that
the r suppression is (most likely) real
the xml files are /ligo/svncommon/SeiSVN/seismic/BSC-ISI/H1/ETMX/Data/ETMX_Performance_060914.xml
and /ligo/svncommon/SeiSVN/seismic/BSC-ISI/H1/ETMY/Data/ETMY_Performance_060814.xml
Sheila has for less gain peaking at the microsiesim in rX and rY and at 60mHz in rZ so i'll make a new set of blends