We made a survey along the tubes in the last 1km arm reaching the terminal buildings. To be reminded, these tunnels are powered with the NEB and WEB IPS mains line, likewise the first slice in the beginning is powered by the CEB IPS, while the intermediate tunnels are conected to the 1500N and 1500W IPS mains line.

We probably found the device generating the +/- 6Hz sidebands of the mains in the WEB magnetometers: it could be a small (2 kVA) UPS installed on December 28, 2023 at 2100m west arm (900m from WEB). This UPS is only connected to the " T " IPS line, unlike other UPSs (2400m) which are three-phase and produce the well known 13 Hz bump with its associated 50 Hz sidebands.

In the attached figure, the top row shows the magnetometers in the experimental room, while the bottom row shows the small triaxial magnetometers near the towers (vacuum chambers). Signals of a large (~8 kVA) three-phase UPS at NEB (blue trace), and those of the small single-phase 2 kVA UPS at WEB (red trace) are clearly visible in the upper left pane.

We will schedule a shutdown to verify this hypothesis.

Today during maintenance we positioned one Bartington 3D magnetometer in the WEB SAS close to the cable trays, at the same height as the bottom tray about 50cm away from the tray. Pictures of the installation are attached. The probe names are:

• ENV_WE_MAG_TEST_X - oriented towards East 
• ENV_WE_MAG_TEST_Y - downwards
• ENV_WE_MAG_TEST_Z - paralllel to cable tray going North

The probes started taking data at about 9:15 UTC. At 9:48 UTC we switched off the lights of the hall. 

Yesterday afternoon we returned to WEB for the scheduled shift on scattered light, which could not take place because of ITF problems.

We profited to move the Bartignton magnetometer in a new location: we placed it on the floor of the SAS, approximately in the center and we kept the same orientation of the axies.  The attached pictures show the installation. The good data are Oct 8 from 16:35 to 17:10 UTC. Same channel names. In this location the magnetometer is about the same distance from the cable trays in the SAS as the Metronix magnetometers are from the cable trays in the hall, while the Bartington is closer to the chillers: see the map in the third attachement. The map provided by Massimo shows in red  the path of electrical cables.
 

In order to investigate a solution to shield the wires creating this magnetic glitches, we performed a few tests on a cable tray piece.

On Wednesday, November 20th, 2024 we taped an untwisted pair of parallel straight wires either directly to the floor of the electronics lab or inside a U-shaped cable tray w/ or w/o its top cover. Note that the cover is not adapted to the bottom part of the cable tray, it is larger and therefore does not fit. The attached pictures show the different configurations tested.

The straight part of wires measures about 3.5-4 m. At one end, the wires were linked by a resistive load (15 Ohms and later 7.5 Ohms) and at the other end an AC voltage was connected.

This setup is equivalent as having a dipole field around the wires, at least in the middle part.

A generator was used to apply an AC voltage at 60 Hz (away from the main power at 50 Hz): Vrms = 10 V which corresponds to about 0.33 A with using the 15 Ohm load.

A Bartignton flux-gate probe was set on its tripod (about 50 cm above the floor) close to the middle of the wires with the down going Y-axis nearly above the wire pairs. The voltages output from the probe for each axis were measured in several conditions and their values are reported in µV  for housekeeping in the attached file data.txt.

The conclusion from these measurements is that there is no shielding effect from the cable tray @ 60 Hz.

We performed further investigations and measured, w/o the cable tray, the field components when moving the probe from one end of the wires to the other end by steps of the tripod width. This was done in order to see if there was not any large field distorsions which could bias our measurements. Everything looks in agreement with what we expect from our setup: Bx increases when moving towards the ends, By stays nearly constant except at ends, Bz increases from the resistive loads towards the generator.

We also replaced the cover by a shorter µ-metal plate and saw no improvement, but this was mainly a test and does not correspond to any possible way of shielding.

For our last measurements we performed a frequency sweep up to 500 Hz with a white noise injection. The current has to be doubled to obtain enough power over all the frequencies range. This was done by using the 7.5 Ohms load.

W/o cable tray: TF ~ 55e-6, flat spectrum

W/ bottom part of cable tray: TF~35e-6, truning point @ -3 dB is 390 Hz

W/ bottom part and top cover: TF~32e-6, turning point @ -3 dB is 320 Hz

It looks like the cable tray material has not only a low permeability, but also a low conductivity and is not able to shield low frequency magnetic fields, unless the length of our sample  piece is too short compare to its aperture. Some simulations could help understanding the possible influence of such an aspect ratio.

Note that the company Beshielding.com proposes cable trays able to shield low frequency magnetic fields.

This entry provides  a more detailed analisys of the acoustic noise emitted by fans cooling the electronics inside the racks at WEB.

A spectrogram reveals the loudest lines emitted at the fundamental acoustic frequency of each fan group
(just a reminder: fans usually operate in the range of 30 - 40 revolution per second, but with 5 or 7 blades, their acoustic emission is 5 or 7 times the fundamental rotation frequency).

This is the elog of "clean data" timing, with loudest lines highlighted in bold.

It is important to note that some of the acoustic emission lines are also clearly visible in the accelerometers placed on the tower vacuum chamber.