Earlier this week, I have installed new symmetrization filters and new damping filters on ISI-BSC8. The damping filters are fed by the geophones signals in the Cartesian basis (L4C for stage 1 and GS13 for stage 2). It works fine (spectra). All filters are built using one unique set of transfer functions (~400TFs). Next, I started to work on the isolation filters. After installing them, I was surprised that the low performance controllers were not all stable. The scripts to build the filters were retested in August 2011 and I didn't work on them until now However, the model was modified.
I compared the transfer functions of the damped ISI measured when dirving in the Cartesian basis with the simulated transfer functions (using TF measured when the ISI is undamped). They don't match as shown in the attachment (Z and RZ seem swapped). Next, I looked carefully into the model and I saw that some connections have been rerouted. In the Cartesian basis, ISI channels are in the following order X, Y, RZ, Z, RX, RY and the HEPI channels are: X, Y, RZ, HP, Z, RX, RY, VP. This order is not respected after the CPS alignment matrices (for stage 1 and stage 2) if we want to have identity matrices by default (no correction). I will fix the model on Tuesday.

Comparison_Measurements_vs_Simulation_TF_ST1_Z_Drive_2012_05_25.fig