When analyzing S3 and S2 data with PowerFlux we see a lot of line artifacts located in bins at multiples of 0.25 Hz. Example plot using ALL S3 H1 data (power in arbitrary units versus frequency):
These artifacts show up in the individual bins only after integrating a lot of data. Not every 0.25 Hz contains a visible artifact.
The following plots are made with a program designed to expose these lines using a single 30-min SFT instead of hundreds of SFTs from entire run. These plots show that the strength of contamination varied during the run and that this contamination can be easily detected in individual 30-min SFTs
For each 30 minute SFT the spectrum between 200Hz and 1500Hz was split into 0.25 Hz bands and the result was averaged - i.e. the frequency spectrum was averaged modulo 0.25 Hz.
Weighted average was used, with weight being the median of a particular component 0.25Hz band.
Bands within 5Hz of 60Hz artifacts were excluded. Also a small number of bands were excluded based on high noise level (these would come from calibration lines for example).
The first column in the table shows plot of bin 0 - i.e. result of the average of frequency bins at 0.25 Hz increments.
The second column shows the same plot but for bin 55 - which can be considered as "good" behaviour.
The last two plots show bin-wise plots for two particular SFTs
| Interferometer | bin 0 | bin 55 (far from 0) |
|---|---|---|
| H1 |  |
 |
| L1 |  |
 |
