Evan Goetz, Ansel Neunzert, Sudhagar Suyamprakasam, Athena Baches, Abby Wang

Summary:
As part of the Jan 2023 DetChar noise sprint, we have investigated narrow spectral artifacts in H1 h(t) data from Nov 2022 - mid-Jan 2023. Several interesting artifacts have been noted, some of which can be attributed to known detector changes, but others have not yet been identified with a known cause or change. We anticipate that some of the artifacts caused by purposeful injection will not affect O4 data.

Methods
Using Fscans linked off of the H1 Summary Pages (for example, Jan 19 2023 (note it is building the page as of the time of this writing), we have searched the daily-averaged and normalized spectra for features automatically tagged as combs and also checked for any visibly noticeable changes in the narrow artifacts. Additional channels are available through the full Fscan navigation button. We have also looked extensively at the interactive plots, which are available but not linked to the main summary pages yet. (Example)

We also looked at some preliminary results from the line history clustering code (experimental, see previous use in this alog alog), which helped us pick out artifacts that vary similarly in time. PDFs attached to this report show November and December cluster plots. Yellow pixels indicate the line was present above threshold, blue indicates below threshold. Each row is the history of the line. Notice that each cluster has similar time evolution.

Key finding 1: The general lines situation since ~January 12

We paid a lot of attention to the vicinity of January 12 as a change point for two reasons: it looked visually different than previous spectra, and it corresponded to a change we had previously been asked to investigate (Robert installed the terminator on ETMY ESD bias cable in the afternoon on January 12). While the key question is easily phrased (did lines get better, worse, or stay the same?) the answer is not. Here's our best summary.

Did something change? Yes.
Is it worse? Maybe.
Is it neither better nor worse? Maybe.
Is it better? No.

Basically, the spectra after Jan 12 look visually different/worse at first glance. However, the primary factors are (a) the very strong 11.0 Hz comb which dominates the averaged spectrum but is actually intermittent, and (b) a variety of other low-frequency lines which are probably not persistent. Fixing the 11.0 Hz comb may enable better insights to answer the questions typically asked.

Key finding 2: A catalog of comb clues

 A strong, intermittent 11.0 Hz comb has appeared frequently since ~January 12.
A clear 11.0 Hz comb is visible in the Fscan data regularly after January 12, and can be observed on the summary pages as well, so it is strong. The comb is occasionally (not regularly) observed as far back as Dec 13. It also clearly turns on and off at certain times. See for example Dec 15 2022 normalized spectra or Jan 19 2023 normalized spectra
We have not determined where this comb is coming from, but suspect that it is human-caused.

The 1.661124 Hz comb (a continuing and important problem) is coherent with several ASC-OMC channels, not just the one originally noted.
Where the comb appears:
H1:ASC-OMC_A_PIT_OUT_DQ
H1:ASC-OMC_A_YAW_OUT_DQ
H1:ASC-OMC_B_PIT_OUT_DQ
H1:ASC-OMC_B_YAW_OUT_DQ

There are also some weak signs of the comb in CS LSCRACK and SUSRACK magnetometers. Example: H1_PEM-CS_MAG_EBAY_LSCRACK_X_DQ

Odd harmonics of an 11.9046 Hz comb in DARM are coherent with many magnetometers, and with ESD voltage monitors
This comb appears across EX, EY, *and* CS magnetometers, and EX and EY ESD voltage monitors.

The 4.98423 Hz comb started around December 8 2022 and is (weakly) coherent with CS EBAY SUSRACK magnetometers.
Although not very strong, it has been regularly present in recent days.

Key finding 3: near-200 Hz lines, calibration line intermixing, and other non-comb features

A strong 201.0 Hz line in Jan data intermixed with low-frequency calibration lines to create a variety of new artifacts.
We correlated this artifact with LHO aLOG 66268 line to measure TCS tuning. Georgia confirms this is intentional and temporary. We notice that the low frequency calibration lines are intermixing with this line creating "sidebands" mirrored around this line.

A strong 200.1 Hz line in Nov/Dec data intermixed with low-frequency calibration lines to create a variety of new artifacts.
This 200.1 Hz artifact was present in November to early December but has since disappeared; note that it is distinct from the strong 201.0 Hz artifact in January described above. Our information about the 201.0 Hz comb (which looks very similar in the spectrum) leads us to suspect that the earlier 200.1 Hz line was also intentional but are not able to find an aLOG describing it. If this is familiar to anyone, please add a comment. The low-frequency calibration lines mix with the 200.1 Hz line creating "sidebands" mirrored around this line.

Check of extra noise from 33.43, 56.39, 77.73, 102.13, 283.91 Hz Pcal lines
We observe these clearly in the data, though we do not yet see evidence for intermixing with other loud lines or each other. But we caution that we have only inspected the daily data and that further investigation is needed to rule out any contamination issues. Meaning, we have not concluded this check.