Virgo Logbook

Environmental Monitoring (Environmental Monitoring)

fiori, paoletti, miller - 14:44 Tuesday 14 June 2011 (29653)

3kHz source, likely in crates of DSP PR and INJ

Today we have "sniffed" around with a magnetic probe trying to localize the source of the ~3kHz noise lines (see also Moving bump around 3kHz caused by switching power supply of rack 39?). After sniffing around potential sources inside the LL, and EE room, and being about to give up ... (!!) we gave a try to the DAQ room, and (luckily) found one 2930Hz and one 3280Hz signal becoming very intense when getting closer and closer to, respectively the DSP crate of PR (2930Hz) and INJ (3280Hz). Figure 1 shows one magnetic spectrum at each crate. In Figure 2, you can visually compare these spectra to the noise in Dark fringe, and other (apparently unrelated) channels like the magnetometer (read out in crate 39) and the Bs_MOD_6Mz signal (generated by the MOD board in LL and read by one ADC in EE room).

It was not possible to locate better the source inside the crate, but we have a few more indications:
- the noise is really from a source oscillating at 3kHz, we can exclude it results from the beating of higher freq. signals, in fact the 3kHz is indeed the largest signal frequency component inside a 2MHz bandwidth (as we measured with another HF coil magnetic probe);
- we can exclude the source is the switching power supplies of the crates (the signal does not get larger as we put the probe very close to them);
- the 3kHz signal seems to get STRONGER when we put the probe close to CAMERA boards (#29 in PR crate, and #31 in INJ crate) or the TIMING boards (next to them): see Picture (Figure 3).

Images attached to this report

Comments to this report:

masserot - 16:47 Tuesday 14 June 2011 (29654)

It seems that this operation trigs some trouble on the VME Timing because the frame stops to be sent and the Tolm packets too:
From FbSusp
Jun14,11-12h07mn14>INFO...-[TolmFrameBuilderSourceManager::Clean] Source IB (version 1) stopped
Jun14,11-12h08mn03>INFO...-[TolmFrameBuilderSourceManager::Clean] Source PR (version 7) stopped

The scan with the sniffer seems to start by the IB crate then the PR crate

Strange state ....

Carbognani, Dattilo, Ruggi - 16:45 Wednesday 15 June 2011 (29666)

Indeed it seems that this monitoring action has create some troubles in the PR and INJ crates which cause is not jet clear. During the troubleshooting all cables on the crates were carefully visually inspected without finding any problem.

Events sequence has been the following:
1) INJ and PR Gx signals disappear from the DAQ
2) Checking in DAQ room we could see that the TOLM on rioprb and rioibb were not blinking in transmission.
3) By restarting the GxS process we could restore the data transmission (also by unplugging/plugging the transceiver on the MuxDemux board channel receiving Gx INJ)
4) Everything seemed fine on PR side and control loops could be closed
5) GxS data on injection were instead not consistent so we decided for a create switch off/on sequence.
6) After the restart everything was fine on INJ side
7) While trying to relock, we could not go over Step 3 and Paolo realized that 4 channels on reference mass coil were not effectively actuating.
8) We decided then to switch off/on also the PR create and this solved all problems.

During PR restart we had a problem of TOLM message header corruption after DSP card download. This should be a completely separate problem and is being investigated at the moment.