Reports 1-1 of 1 Clear search Modify search
AdV-DET (Local controls for all benches)
bonnand, masserot. - 14:53 Friday 27 January 2023 (58589) Print this report
Noise in SWEB control

This morning, alain realized that the SWEB local controls shows much higher noise than usual, this can be seen in the vim plot, see figure 1 where we see that the noise in the controls of the angular d.o.f. increased a lot between 22:00 and 23:00 UTC.

Figure 2 shows that it happens roughly wednesday 25th of January at 22:37 UTC.

Figure 3 shows the spectra of the angular d.o.f. and their correction before and after the increase of noise.

We looked at the raw read-out signals of SWEB before and after the increase of noise, see figure 4, we see that the noise around 10 kHz was already high but it increased even more resulting in a reduction of the SNR for the demodulation of the lines around 10 KHz that probably causes the increased of noise.

We don't have clear explanation for that behaviour or what could have caused the noise increase last wednesday, one could think that it could come from the "Nikhef box" used to read-out the LVDT signals.


We then checked all other benches :

  •  SDB1, figure 5, all signals looks OK.
  •  SQB1, figure 6, all signals looks OK.
  •  SDB2, figure 7, all signals have a very high noise above 200 Hz and even the noise floor is higher than on other bench. To be noted that the noise in the local controls is similarly low to other bench.
  •  SIB2, figure 8, all signals looks OK.
  •  SNEB, figure 9, all signals looks OK.
  •  SPRB, figure 10, all signals looks OK except FL_V that shows higher noise between 1-6 KHz.
  •  SQB2, figure 11, all signals looks OK except BR_V that shows higher noise between 0.5  and 8 KHz.
Images attached to this report
Comments to this report:
bulten - 17:13 Friday 27 January 2023 (58594) Print this report
I tried to see if I could see anything strange on the LVDT signals from the SWEB DAQ box; we do make a 10-kHz (or maybe 20 kHz, I have to check what we currently use) pilot signal.
The LVDT and blended signals for SWEB look more or less the same on Jan 25, 22:00 UTC and 22:40 UTC, although the z-position shifted by about 15 micrometers.
The very-low frequenci position oscillations are somehwat larger after 22:40, but still small. Also the LVDT-H2 noise floor is lower before 22:40; I will check the trillium.
For the SWEB d.o.f. we only store signals to 500Hz in the raw file so I cannot look for a problem at 10-20 kHz directly in the raw file.

I think that the fastest solution to identify a possible problem with our DAQ box is to switch off the SWEB DAQ box (open the position PID loop first) and check the LVDT raw signals on
SWEB_LC. If the noise is not modified, then the problem is not coming from our DAQ box. if it IS coming from our DAQ box then I suspect a broken wire making contact inside SWEB:
a wire going to the LVDT driver coil could maybe make some contact to some ground plane. Mind, I would expect a HUGE result then for the LVDT signal involved, and I do not see that
in the 0-500 Hz plots below.
Non-image files attached to this comment
bonnand, masserot - 17:30 Monday 30 January 2023 (58627) Print this report

The noise disappeared last saturday the 28th of January at around 13:21 UTC.

There is no visible action at that moment.

Figure 1 shows the trend of the error signals of the local controls of the bench, ensquared in green are the Horizontal channels and in orange the vertical ones.

Figure 2 and figure 3 show the FFT before (purple) and after (blue) the noise disapeared, figure 3 is a zoom of figure 2 around 10 kHz.

To be noted that the noise floor around 10 kHz is lower than before the excess noise appeared the 25th of January.


The situation didn't change for SDB2.

Images attached to this comment
bonnand - 11:20 Wednesday 01 February 2023 (58664) Print this report

The excess of noise in the SWEB local control is appearing again this morning.

Figure 1 shows the FFT of the raw read-out signals, in blue yesterday (no excess of noise) in purple this morning. Already yesterday the noise was higher around 10 kHz (compared to sunday for example) but with no impact on control noise.

Figure 2 shows the rms of the same signal cut at 1000 Hz where we can see the excess of noise that is propagated to the control signal at low frequencies.

Is there any activities in the SWEB building that could explain this ?

Images attached to this comment
bonnand - 15:39 Wednesday 01 February 2023 (58670) Print this report

Here is a spectrogram (fftime) of 2 LVDTs of SWEB showing what is happening between 2:00 UTC and 8:00 UTC, figure 1.

One can see that there is a peak between 12.5 and 13 kHz which is drifting slowly and then dropping in frequency and become very broad around the demodulation lines ( the very thin lines around 10 kHz).

The coil driver (what I was calling "Nikhef Box" earlier) is probably at the origin of this as it is designed to excite the signal around 10 kHz.


Figure 2 is the same plot at the time where the noise disapeared last saturday (the 28th of January), there the excitation around 10 kHz totally disapear.


First thing we should do is to check the plugging of all the wires on this Coil Driver Box (including cables coming from the bench and power supply cable) which is located in a rack at the west end building.


Images attached to this comment
Search Help