After the switch-off of some vacuum sensors around the WEB tower (see entry 47985) that has "cleaned" a lot the magnetic environment, we repeated the search in a more controlled way (and now with the ITF locked in LN3-SQZ) of a source (better: a "trigger") of the 48Hz line (we suspect it is a mechanical resonant mode of the F7 crossbar) and of other lines located between 40 and 44Hz that are visible in Hrec.
The test consisted in a sequential switch-off of the two UPS systems (they are in parallel) that are supplying all the mains power to the electronic systems around the WEB tower.
Timing (UTC) is:
before 13:49:00 UTC both UPS are ON, as usual
13:49:00 UPS DX OFF
15:14:50 UPS DX ON
15:19:00 UPS SX OFF
16:45:00 UPS SX ON
after 16:45:00 both UPS are again ON, as usual
The clear effect is that the 48Hz line and quite all the lines between 40 and 44Hz disappeared if ONE of the two UPS is OFF (not important which of the two).
It is likely that when they are working in parallel the global noise on the UPS line increase, even if we do not see (or we are not able to see) any clear increase of noise on the UPS monitors and/or in the WEB magnetometers.
Attached many plots showing this effect :-)
It remains unclear the path (who is propagating this noise, using wich path, acting how and where ...), but the effect is visible without any doubt.
A question remains open: is this type of problem present in other places as well? (e.g. NEB, CEB)
Should we do the same shutdown test as (many) UPS systems?
To be investigated thoroughly ...
p.s. it seems that even from 50 to 55hz there is a faint improvement
