After the recovery I did a quick test for PD1 (which is the one in loop). I found that the phase noise subtraction was already on (CARM_NULL_1F from 17:00:31 UTC), then at 17:12:11 UTC I turned on the amplitude noise subtractions and at 17:14:33 UTC I turned off the phase noise subtraction. As seen in the plot there is no clear improvement, at least in CARM_NULL_1F (purple trace only phase noise subtraction on, red trace both on, blue trace only amplitude noise subtraction on).

I left both on (on PD1 only), and added in the automation the disabling of all four (so also for PD2) in DOWN, and the enabling of all four in LOCKING_CARM_NULL_1F (quite before we close the CARM_SLOW loop).

The first plot show the ASD of the RFC_PD1_6MHz channels with the ITF at step 100 :

• the I and the magnetude noise signals and their coherency
• the Q an the phase noise signals and their coherency
• the 6MHz and 12MHz magnetudes and their coherency
• in the red rectangle, the lowpass ratio(6MHz_12MHz_mag_slope) between the 6MHz and 12MHz magnetude and the  full bandwith one (zz_6MHz_12MHz) :
  • at this step 100,  the RFC_PD1_6MHz_12MHz_mag_slope seems to be over-estimate du to the low-pass filtering

The second plot show the trend  of 2 lock acquisition sequences

• one lock acquistion with a trial to engage  the RFC_PD1_6MHz demodulation noises mitigation at the begining of the step 100
  • one can be seen that the noise attenuation of demodulation was initiated when there are still huge fluctuations in the magnitudes 6MHz and 12MHz
• one lock acquistion without  the RFC_PD1_6Mhz demodualtion noises mitigation
  • one can see  that the 6MHz and 12MHz  magnitudes are stabilized when the CARM_SLOW loop is engaged

So maybe the RFC_PD1_6MHz demodulation noises mitigation has been started to early and should be performed after closing the CARM_SLOW loop.

During the lock acquisition , there is a period of 2mn where where the CARM loop is disabled and the CARM_SLOW no yet engaged.

The tests of the RFC_PD1_6MHz demodulation noises mitigation where done during this period (see the plot)

Some tests were made this morning in LN3_SQZ , on the RFC_PD1_6MHz channels with the following conditions

•  2024-10-25-07h28m00-UTC : demodulation noises mitigation OFF , RFC_PD1_6MHz_12MHz_mag_slope as the ratio between  RFC_PD1_6MHz_mag and RFC_PD1_12MHz_mag filtered by  2nd low-passe @ 0.1Hz (plot)
•  2024-10-25-08h00m06-UTC : demodulation noises mitigation ON,
  • 2024-10-25-08h00m06-UTC: RFC_PD1_6MHz_12MHz_mag_slope  filtered by  2nd low-passe @ 0.1Hz (plot)
  • 2024-10-25-08h05m26-UTC: RFC_PD1_6MHz_12MHz_mag_slope  filtered by  2nd low-passe @ 1Hz (plot)
  • 2024-10-25-08h10m41-UTC: RFC_PD1_6MHz_12MHz_mag_slope  filtered by  2nd low-passe @ 4Hz (plot)
• 2014-10-25-08h15m17: demodulation noises mitigation OFF
• Note that
  • the channel RFC_PD1_6MHz_12MHz_mag_slope is stored only at 1Hz by decimation
  • the channel zz_ratio_6MHz_12MHz  is the  same as the  RFC_PD1_6MHz_12MHz_mag_slope channel but at 10KHz

As already reported by Diego, no improvement was observed even with the increase in frequency of low pass filter cut-off applied on the channel RFC_PD1_6MHz_12MHz_mag_slope (plot)

This is an entry written a month ago that I have found in my drafts. I am posting it again in case it will be useful in the future, but I am not sure it is relevant anymore.

The noise seen on RFC 6MHz I seems not to be demodulation amplitude noise caused by large low frequency fluctuations of the RFC 6MHz signal.

Figure 1. Looking at the noise subtraction correction, it is almost a factor 10 below the the 6MHz I noise. But that does seem appropriate, the noise level on RFC 12MHz mag and on RFC 6MHz I are the same, but the total RMS of these signals are different. It is on average higher by a factor few on 12MHz mag. So if the noise on RFC 6MHz I was caused by demodulation amplitude noise, it should actually be a factor few lower.

I have also tried to do an amplitude noise subtraction off-line in matlab and I arrive at the same result.

There are at least two options remaining to explain this noise RFC 6MHz I. It is demodulation amplitude noise, but the RMS is dominated by fluctuations at high frequency (above 5kHz) that we cannot see in these signals. Or it is a noise of a different nature, that has a 1/sqrt(f) shape.