In order to perform the optical losses measurements at WE around 8h-10h UTC, we have stopped the WE PCal permanent lines. During these measurements, we have seen glitches on the power sensors.
We initially though it may be related to the SAT activities to install a PDU on a rack close to the calibration rack, and in particular related to switching on the motor drivers plugged on this PDU. However, the conclusions are not clear finally.
Here are some plots as a reminder:
- figure 1: a short period with glitches during ~3 seconds. The peak-to-peak values of the power during the period without noise are 1.299 to 1.301 W , it increases up to 1.295 to 1.305 W in the noise period.
- figure2: zoom on the same period.
- figure 3: another period, of about 1.5 minutes, with glitch rate increasing, and then they suddenly stopped.
- figure 4: around 9h14 UTC, the glitches seemed to appear at the time the SAT team started the motors. This very glitchy period lasted all along the motor tests, but did not stop at the end.
- Around 9h48-9h54, we restarted the PCal processes and the glitches were not there after restart.
- Figure 5: At 9h56, we made a test: SAT team restarted the motor drivers, and glitches re-appeared with increasing rate. They were stop 1 or 2 minutes later, and the glitch rate decreased and suddenly they stopped.... however, glitches restarted a bit later, without particular action from SAT.
- FIgure 6: summary of the morning activities at WE from the trend of the channels, where the presence or absence of glitches is well visible looking at the min/max values.
- SAT team left WE building around 10h20-10h30.
- 10h30-10h40: stopped all PCal hardware (switch off the Hameg power supply and the DaqBox, ), and restarted all, to compare the noise of the WE and of the NE sphere until 11h30 (see previous logbook entry).
- Figure 7 : at 11h40, put back the permanent lines. The amplitude of the lines is much larger than the glitches (the bottom plot has been zoomed on y-axis to see some glitches before the calibration lines were restarted).
We did not notice such glitches during the run. However, they may have been present, but not visible since the calibration lines were always on.
In the data from this morning, it looks like the glitch rate was higher and over longer periods when there were activities around the calibration and SAT motor racks in the WE building, but it is not very conclusive.
The PCal power loop is using the single-ended output of the fastDAC. It is known that (at least some) fastDAC channels have some oscillation at ~MHz high frequency. A past quick test at NE (logbook 66132) to use differential output was not useful about the noise level. But could it generate these small glitches?
To check if the situation is similar for NE PCal, it will be worth stopping the NE PCal lines also for a long period during the WE mirror replacement next week.