This spectral noise structure seems to be electric in origin. I have run a Bruco on the data from February 12 at 19:00 UTC: link to results. A non-negligible coherence is present at 42 Hz with ENV_CEB_{UPS,IPS}_{CURR,VOLT}_..., and with various magnetometers on the BS, DT, IB, etc. No coherence has been highlighted on the "sidebands" at +-400 mHz from 42 Hz. Interestingly, other small coherences are present at integer frequencies between ~30 and 50 Hz. These might be related to the noisy chiller described by Alain in this entry and its comments: #63274.

Figure 1: spectrogram of Hrec over 7 hours of data around the region of the 42 Hz structure. In fact, two sidebands are visible at approximately +-400 mHz. These may be due to some non-linear coupling, which is why they have not been caught by a coherence analysis. The 42 Hz line seems to move a little bit in frequency, as typical of the 50 Hz electric mains.

Figures 2 and 3 are coherence spectrograms between Hrec and ENV_CEB_IPS_CURR_R_2000Hz and ENV_BS_MAG_W, respectively. In these colormaps, some coherence is clearly visible at 42 Hz and, fainter, at 41, 43 and 44 Hz.

Figure 4: examining three different periods, Dec 25, Feb 12 (no Ncal), and Feb 14 (Ncal back), it seems that this excess noise at 42 Hz was not present during the mini-Engineering Run. Still, some structures are visible at 42.4 Hz (but not at 41.6 Hz). Concurrently, as shown by daily BruCo results for the ENV channels, no coherence with the above CURR and MAG channels was present during the days of the mini-Engineering Run at integer frequencies. This seems to confirm that the comb has appeared late (following the hint in #63274).