The up and down coils of NE and WE are used for locking, so the NE and WE correction (Sc_NE/WE_zDAC_UD1) is sent to them. Instead left and right coils are not used. Indeed the first plot shows that up and down contain some correction, while left and right do not.
The second plot shows what Irene already found: there is very good coherence between all NE coil currents and dark fringe, and very small coherence of WE up and down coil currents with dark fringe.
Since no correction is sent to NE left and right coils, the coherence with those currents means that the origin of the bump must be traced down to those coils.
The situation might be a bit more complex for up and down, since the bump could be present there because it is re-introduced by the correction. To rule out this possibility, I measured the transfer function from z correction to NE and WE up and down coils, while injecting DARM noise. I then used it to project the current expected on the coils as an effect of the correction. Plots 3,4,5 and 6 shows such projection for all up and down coils. It is clear that the bump visible on currents is not the result of the correction.
In conclusion, it seems very likely that the 88Hz noise bump is generated by some real noise present on all NE coils that produces a real motion of NE mirror.
