At about 9:50 UTC up to 9:55 UTC we switched off both UPS machines. Automatically the UPS load (racks etc.) switched to the IPS power line. The result is visible in the first plot, which shows the noise of the UPS line (purple when still under UPS, black when switched to IPS). The reduction in noise is about a factor 100. The second plot shows the reduction of magnetic field in the central hall measured by magnetometers (purple when still under UPS, black when switched to IPS). The magnetic noise reduced about a factor 20 between 0 and 1000Hz. Figure 3 shows a spectrogram of one magnetometer while switching from UPS to IPS.
Answering the question posed in eLog 25297, we now know that the noise in the UPS line is not produced by our loads, but it is intrinsic to the UPS machines.
To be noticed that the UPS machines of the MC, WE and NE (of different model and smaller size) are less noisy: see UPS sniffers moved.
