The models in this section have evolved from the original GEO triple model by Calum Torrie, except that in more recent versions the original hand-derived state-space matrices have been replaced by matrices exported from the Mathematica models, to reduce errors and incorporate more physics. They all break the full system down into four independent subsystems: longitudinal/pitch, transverse/roll, yaw and vertical. Thus they run quickly in time-domain simulation but neglect any asymmetries. They are all self-contained - download and unpack them anywhere and consult the help file (ahelp_xxxxxx.m or similar) for further instructions.
12/13/06: 20061213_T060283-02 (a.k.a. T060284-01). Suite of models for ITM/ETM, compensation plate, and ETM reaction mass using parameters from T060283-02. This package can also be found on the DCC as T060284-01.
09/14/06: 20060914quadnoise.zip - Third cut at the noise prototype, with additional numbers from T010103-05. Corrects an error in the derivation of wirelengths in 20060517quadnoise.
08/25/06: 20060804BS.zip - A proposal for the AdvLIGO beamsplitter, with only two blades at the top mass rather than four as traditionally.
Note: this model is significant not only for the particular parameters but because the Matlab code for generating state-space matrices has been revised and extended to incorporate many features from the quad models.
It loads one of four different sets of programmatically generated matrix elements (symbexport3.m, symbexport3db.m, symbexport3dblat.m or symbexport3lat.m) depending on the selected options.
If you set pend.db=1 as in the example, you get the new variant with dual blades at the top mass. If you leave it undefined or 0, you get the old GEO style.
If you set both pend.kx1 and pend.kx2 to per-side lateral stiffnesses, you get laterally flexible blades - otherwise you get ideally laterally stiff blades.
From the quad model, if you set pend.stage2=1, it treats d0-d4 as raw attachment point positions and adds the effect of wire stiffness separately. If you leave pend.stage2 undefined or 0, it assumes d0-d4 are effective values with any flexure correction already applied.
From the quad model, if you set pend.ribbon=1, it applies fudges to approximate ribbons.
05/22/06: 20060517quadnoise.zip - Third cut at the noise prototype, with additional numbers from T010103-05.
05/13/06: 20060509quadcontrolsmainasbuiltsym.zip - The as-built quad controls prototype main chain. It corresponds to case 20060509controlsmainasbuiltsym of the quad xtra-lite lateral model. It (necessarily) neglects the off-axis terms in the MOI tensor due to the asymmetrical top and upper intermediate masses.