This morning we've investigated about it.
Some remarks:
1) the noise is usually visible in Em_RFBDCE_tmp receiving antenna as a "bump" located ABOVE the VIRGO 6.263xxx MHz modulation frequency.
This bump is not stable in frequency, but drifts slowly in time. Moreover it has some fast "glitches" toward a lower frequency (some kHz below) with a slow recovery toward the previous frequency; the amplitude is quite constant.
2) the same noise is visible in B1 photodiode (more specifically in Pr_B1_d2 but NOT in Pr_B1_d3), even when the interferometer is not locked. It is also visible in some other photodiodes (e.g. Pr_B2), but not in all of them.
3) A wider analysis on the RF antenna signals revealed that this bump is part of a "family" that is visible every 156.6kHz, starting from about 1MHz up to slightly above 7MHz
4) using a sniffer (Agilent E.M. probe connected to a Spectrum Analyser) this noise can be detected over many cables in DAQ (maybe more on timing cables, but it was not easy to judge); anyhow it seems present everywhere. We've tried to locate it with a tuned receiver, but due to its nearby with the VIRGO L.O. frequency it was impossible to disentangle (receiving window was about 5kHz).
Taking the opportunity of a VIRGO unlock, we've tried to swap some cables in Detection Lab, trying to understand the different photodiodes behavior.
5) swapping L.O. cables between d2 and d3 Demodulation boards nothing changed (bump noise remained on d2)
6) swapping RF cables between d2 and d3 (on demodulation boards) moved the noise from d2 to d3
7) swapping RF cables between d2 and d3 (on photodiodes boxes) moved the noise from d2 to d3
8) swapping power supply cables between d2 and d3 (on photodiodes boxes) moved the noise bump from d2 to d3 but some noise remained on d2 (not evident result)
9) finally clamping ferrite beads over power supply cables (near photodiode boxes) removed at all the bump (while ferrites on RF cables didn't produced a noticeable result).
Preliminary conclusions:
- There is a source somewhere in the VIRGO Central Building that create lines spaced 156.6kHz between them; these lines are running over many cables. This frequency could be a Switching Power Supply, could be a Serial Digital Line, or something related to Cameras (15625x10 = about 156.6kHz) etc. Whatever it will be, it is something not stable in frequency, with random dips in frequency and with fast transitions (is not a sine wave).
- This source couples mainly with photodiodes power supply cables in a "common mode" way (a ferrite is able to reduce it).
First attached plot compares three different periods: blue this morning before ferrites, purple today after ferrites, black past week when the bump was at high frequency. The noise is still visible in the RF antenna, but no more in the B1 diodes.
The second attached plot shows how the coherence between RF antenna and diodes reduced a lot. Black is before ferrites, green after ferrites.
This problem was already solved at WE in the past putting ferrites on B8 power supply cables and then by moving the temperature box (see entries 8365, 10435).
In the central building the culprit was more difficult to identify since there are many temperature sensor boxes, but was solved once (entries 13227, 15556). Probably the noise is present in all temperature boxes but jumps from one to another (since its frequency is temperature dependent as underlined in 13195). This kind of temperature sensor boxes should be avoided for AdV.