Since some days we have a sensor (ENV_CEB_ELECTRIC) that is no more working (saturated channel). This sensor detect and amplify the voltage drop between the Vacuum Tube pipe (metallic structure just below the WI tower) and the earth safety system (metallic structure of the WI Sc crate, just over the platform, near WI tower).

in the past we used this signal to monitor and find the 50Hz sidebands, and it is a good witness for us; so yesterday, during the maintenance, I've been in CEB with a spare electronics (Stanford SR560 preamplifier), but the problem is not in the preamplifier.

It seems instead a "common mode" problem: if the + and - input are tied togheter and left floating, the output is near zero; but if they (tied togheter) are put at ground (the vacuum pipe) the system saturate, like if the point where we connect it would be outside of the common mode of the preamplifier. Or, saying it in another way, like if *the preamplifier* is floating (example: if the earth safety connection inside the mains plug has an offset that brings the preamplifier out of the working point).

Clearly this is impossible (Vacuum pipe and safety earth are at the same common point); but if there is a high frequeny noise problem, the earth connections have their inductance, so this last statement is no more true.

Indeed looking at some other channels at the time when the channel ENV_CEB_ELECTRIC started to saturate (October 8 at 22:17:00 UTC) I've found that also other signals become unusable (e.g. ENV_IB_FINGER_ACC_Y), and the RF antenna start to detect a higer RF level (peaks at 200kHz and 2MHz, with also a broadband noise).

This morning we have been in CEB in search of some obvious noisy source (e.g. some broken/defective switching power supply), but in the quick search limited to the EE_room and INJ tower location we have not found any evident culprit. Situation remain as it is for now, waiting for a more detailed search taking advantage of the next maintenance period.

Attached: some plots related to signals before and after Oct. 8

The effect of this noise is visible in many sensors, all of the kind ENV_xxx_ACC_xxx (see first and second plots).

Is not visible in other families of sensors (e.g. _MIC_ or _SEIS_ or _MAG_) apart a little increase of background noise in the ENV_CEB_IPS_VOLT family

The RF seems a ~317kHz (+/- 6kHz) line, with harmonics, that is not constant in time (see the attached spectrogram)

Could it be some switching power supply that started to misbehave ?

We looked all inside the Central Building with a 3-axis portable magnetic and electric spectrum analyzer, and eventually found a small switching power supply that was the "noise generator" since October the 8th.
Switching it off has restored the original situation in the RF antenna, in the ENV_CEB_ELECTRIC signal and in many accelerometers.

The power supply was plugged to a scanning Fabry-Perot that we left out for the moment, we will look for a new power supply as soon as possible.