Displaying report 1-1 of 1.
Reports until 16:19, Thursday 10 July 2014
H1 SUS
brett.shapiro@LIGO.ORG - posted 16:19, Thursday 10 July 2014 (12693)
QUAD08 model fitting results and wireloop model updates
***********************************************************
QUAD08 model fitting

I ran the model fitting algorithm on QUAD08 to see if any more insight could be gleaned from the measured TFs towards what is causing the error in the 2nd pitch mode. The fitting algorithm is on the svn at:
.../SusSVN/sus/trunk/QUAD/Common/MatlabTools/QuadModel_Fit/QuadPend_GaussNewton_fit_v3_X1QUAD08.m

The results confirm what others have already seen, in that increasing dm or dn (the top mass wire flexure heights relative to the center of mass) makes the model match the measured pitch modes. Specifically, the model says that either

+1.4 mm on dm

or +1.7 mm on dn

equivalently brings the model in line with the measurements. You can also trade offset between one and the other because these two parameters are degenerate in the model. I also changed

h1 = 0

to 

h1 = 2 mm.

h1 is a horizontal front-back shift in the center of mass for the UIM stage. Betsy and Arnaud observed that the pitch adjusters here are pushed far forwards. This parameter reproduces much of the vertical to pitch cross coupling seen the the vertical-vertical and pitch-pitch TFs. It does not influence the mode frequencies.

I put a QUAD08-tuned model on the svn here:
.../SusSVN/sus/trunk/QUAD/Common/MatlabTools/QuadModel_Production/quadopt_wireloop_QUAD08.m


The physical cause of the increased pitch frequencies remains a mystery. Drooping top mass blades would do this, as would improperly seated wire clamps for the top most wires. Since the top mass blades have been replaced without altering the pitch modes, it is unlikely to be those. A 0.15 mm increase in the top wire diameter would also do this, but Arnaud and Betsy checked the wire with calipers and it is correct.


***********************************************************
wireloop model updates


In the process of doing this analysis, I updated the wireloop model on the svn, which was quite out of date. See:
.../SusSVN/sus/trunk/QUAD/Common/MatlabTools/QuadModel_Production/quadopt_wireloop.m

Based on the most recent drawings for the wireloop quad (all metal, wireloop under PUM) I updated
* all mass and inertia values
* 3 of 4 wire lengths: ln, l1, l2  (these were based mostly on previous work with Mark Barton on the fiber quad model updates)
* transverse wire spacings: n3, n4, n5
* and the d values: d2, d3, and d4

The adjustments on the n parameters helped bring roll more in line. The pitch inertias helped pitch a bit as well. The other parameter corrections had little influence on the dynamics, but are at least more representative of the as-built system.

This work also found an error in some of the assembly drawings where the PUM wire clamp was flipped upside down. This has now been corrected in the drawings. The clamps had been installed correctly.


***********************************************************
The attachment shows a comparison of TFs of the original wireloop model, the new wireloop model, the QUAD08-tuned wireloop model, and the QUAD08 measured TFs.
Non-image files attached to this report
Displaying report 1-1 of 1.