This morning Paolo made some noise injections on the angular controls of SIB1. He prepared a noise shaped like in figure 1 (4 zeros, 2 poles @ 20Hz and unitary gain at 100Hz with amplitude chosen from the DSP card, different for each dof). This noise has been injected on the marionetta corrections (Sc_IB_MAR_TX,TY,TZ_CORR channels).
GPS start | DOF | dur (s) |
1264933968 | IB MAR TX | 240 |
1264934448 | IB MAR TZ | 240 |
1294935274 | IB MAR TY | 160 |
The quiet reference was chosen at gps 1264932318. The plot 2,3, and 4 are the coherence with DARM and the FFTs of the IMC automatic alignment error signals for each injection, showing the resonances which are excited.
We then computed the transfer functions of the marionetta corrections with Hrec and performed the projection on the quiet gps the point having coherence greater than 0.4.
The transfer functions and coherences are are shown on fig 5, while the projection on Hrec on fig 6.
We actually should add to the marionetta corrections the model for the coils, otherwise we risk to underestimate their contribution. We'll include it as soon as possible.
Another remark is that the TY degree of freedom doesn't show coherence except for the 19 Hz, where it is already noisy even when no noise is injected. We sohuld better understand why the projection for this didn't work (it is ~10 times above Hrec), and we also should perform again the injection, with more noise on TY, and see what happens in that region.