I've investigated the bump between 1.05 and 1.12 kHz by estimating the coherence of the channel V1:SDB2_B1_PD1_Audio_100KHz with all the raw channels with sampling rates larger than 20 kHz (46).
Figure 1: the 6 channels with significant coherence at the frequencies of the noise bump are shown in this figure, which represents the coherence with the target channel in the band 10-15 kHz. Some channels are clearly interdependent and naturally correlated with the target. I let the experts comment on how informative this result can be. Notice that the curves for V1:SSFS_Corr_50kHz and V1:SSFS_Err_*_50kHz are superimposed.
Figure 2-3: noise projections with some of the above most coherent channels (ASD vs frequency: sorry for the missing axis labels)
I repeated the same analysis with all the raw_full channels with a sampling rate larger than 20 kHz (247):
Figure 4: the only channels exhibiting a significant coherence with this peak in SDB2_B1_PD1 are from the V1:EDB_B1s_QD1_RF_56MHz_* family. Not shown, a small barely significant coherence has also been measured with V1:SNEB_LC_LVDT_FL_V_out_raw_DS and the V1:SIB2_LC_LVDT_*_out_raw_DS family.
Comments and suggestions are welcome on whether these results are informative and what to test next.