We finally have a good explanation for the oddities we saw in the West End tower.

As we have known since a long time, mirrors are "charged"; when an external electrical noise is brought inside the tower, down in front of the mirror, it applies a force that transfers the noise into the Hrec.

       Timeline and details

For all the tests up to the end of January, read this document.

Monday 27 Jan 2020 we went back at WEB trying to understand what happens leaving both mains wires (hot and neutral) opened in the motors crate: we did it using the main switch on the front panel which, contrary to the "smart-plug" to which the motor crate is plugged in, it disconnect both wires of the mains, not just one of the two. All the noise that we used to have in Hrec has disappeared, in the same way as when we disconnect the cables of the motors going inside the tower, or pulling out from the motors crate the mains plug. This is the logfile.

Thursday 30 Jan 2020 in the morning: during an unlock we went into CEB, and we opened both mains wires of all the motors crates like done in WEB. It seems some lines in the usual 40-50Hz region were reduced, but we did not have the same global noise reduction obtained at WEB. This is the logfile.

Thursday 30 Jan 2020 in the afternoon: we again went at WEB with a low frequency signal generator, and we applied a 100mV 15Hz signal on the mains socket of the motors crate #7 (the one to which the two most offending motors cables are connected: MH.MA.Tx and MH.MA.Tz). The reason of that operation is that these two cables (and only these two) runs togheter into the tower with the coils cables. At the beginning we hade some troubles (signals not visible in Hrec and ITF unlocked for an earthquake), but we later were able to clearly see lines growing in Hrec. A line at injected into the mains socket (1V, 15Hz) with the front panel switch ON and the motors cables connected, is visible at f and 3f in Hrec (and maybe also at 2f and at higher orders). The amplitude of the fundamental line increase with the square of the applied level. Moreover when we injected a line in the 48Hz region (49.44Hz) we excited many other lines all around (but the Paolo's dampers are ON, so this test has to be repeated in a more controlled way). This is the logfile.

       Conclusions

What we see is well explained with this theory:

1. we have the mirror that is "charged", and an electric voltage applied on it is seen in Hrec.
2. the ground/earth system in Virgo is very good, and the towers body is rock-steady grounded ad zero potential, as well as the racks and crates
3. the mains wires (hot and neutral) have resistance and inductance, and loads asking current create voltage drops (and this is normal)
4. this voltage drops (the noise) is transferred from the mains to the motors cables that goes inside the towers as a "common mode" noise. To all the motors cables. We don't know why (ask to the group that designed motors crates a deeper analisys)
5. two of these motors cables runs inside the towers for tens of meters in a bundle of wires that has also the coils cables, transferring noise to them in a capacitive way
6. the coils acts on the mirror (not on magnets! No current is flowing into the coils) transferring the noise in Hrec

It remains to understand why we didn't see a relevant effect in NEB also: this could be due to the charge of the mirror (not the same?) or because the loads on the mains wires are not the same in WEB and in NEB (less loads and less noise in NEB), or maybe the cables shield is broken somewhere along the chain. More analisys is needed, e.g. redoing this signal generator test at NEB and so on.

Possible action for solving this problem

• discharge the mirrors :-)
• avoid transferring mains noise from crates mains socket to motors cables (modify/redesign motors crates?)
• use double-pole smart-plugs, opening BOTH wires of the mains supply
• leave crates master switch OFF (these motors are seldom used); or leave these two (and only two per tower) motors cables unplugged from the crate: MH.MA.Tx and MH.MA.Tz (to be checked if they are always the same in all towers)

We plan to perform other tests and repeat these tests on other towers (NEB as first one), in order to have a better understanding of all this behavior.

We still emphasize the fact that THIS IS NOT A "GROUND PROBLEM".

Attached: some plots and spectrograms of the last tests done in the afternoon. We left the WEB motors crate #7 with the master switch OFF and many motors cables unplugged, we will come back in a stable configurration at the end of the tests.