Today we tested the lock acquisition procedure integrating again the 6 MHz RAMS Servo along the modulation path, as planned.
We did it in two steps: first we kept the Servo loop open (~9:00UTC) with the goal of making sure there is nothing in the way the electronics is built that prevents completing the lock acquisition, belief that had gained momentum due to previous troubles.
After retuning the phase as successfully done during the Noise Injections of last December (#62704, #62783, #62814) and a couple of unrelated unlocks, we smoothly reached Low Noise 3.
Then we closed the servo loop (~10:50UTC) and repeated the whole procedure, again reaching low Noise 3 without complications.
At this point since we still had a couple of hours before the end of the shift, we decided to use the opportunity to try and test the (original) 56 MHz RAMS Servo as well, unmodified. This time we went straight for the jugular and did it with the loop closed (~13:25UTC). Once again, everything went smoothly and we managed to reach Low Noise 3.
We left the two servo on and online, with the monitoring channels (useful to evaluate performance of the servoes) acquired. To be noted, the 56MHz monitoring channels have been plugged after daily meeting at ~15:00 UTC.
In the plot attached you can see the sensitivity in LN3 measured yesterday (without servo) today with the 6 off and on, and finally with the 56 on.
No effect of the suppression is visible with the current level of interferometer sensitivity.
What is really surprising is that there is no effect on the audio channels of B1s, B5, B4, B2. Back in July 2023 we had measured that the spectrum of these signals move up and down with the 6MHz and 56MHz modulation depth
Figure 1 shows that there is no difference between before and after the RAMS installation yesterday
FIgure 2 There was a test of the RAMS back in April 2023, with two lucky locks with the RAMS enabled. In blue is Apr 19 with the RAMS enabled, and in red Apr 18 without the RAMS installed, and for reference in purple is the situation from yesterday. Back in April the RAMS had an effect, the improvement was small, but at that time the 6MHz and 8MHz sidebands were cross contaminating each other with the 2MHz laser noise.
There are two hypothesis that I can see
- The audio channels of the B1s, B5, B4, B2 are seeing noise in some other property of the sidebands than RAM. Despite that back in O3 the level on B1s was exactly at what is expected based on the measured electronic RAM level and the theoretical filtering of the interferometer.
- The RAMS is not functioning.
To check hypothesis 2 there is a simple direct test can be done. Lock the OMC in single bounce on the 56MHz sideband (the one below the TEM00 to be far from the carrier order 1 mode), and turn on and off the 56MHz RAMS servo loop. During the RAM test of July, the effect of the RAMS servo was clearly visible.
Below some plots comparing some channels with the 6MHz RAMS
- 2024-01-10-10h15-UTC : ITF_LOW_NOISE3 - 6MHz RAMS loop open (purple)
- 2024-01-10-12h00-UTC: ITF_LOW_NOISE3 - 6MHz RAMS loop close (blue)
The first plot ans its zoom show the coherency betweeen the B2,B4.B1p,B1s Audio channes, the B1_DC one and Hrec
- at 2024-01-10-12h00-UTC, it seems that the 6MHz RAMS noise is greater that the one at 2024-01-10-10h15-UTC .
- Even if the 6MHz RAMS loop is closed, the coherence with the B2 and B4 Audio channels increased
- the coherence with the B1_DC channel remains at the same small level
- While for the one with Hrec seems to increase a little bit in the 10-30Hz frequency band
The goal of yesterday's test was to make sure there is nothing that prevents the completion of the lock acquisition procedure when using the RAMS Servo electronics.
Based on previous attempts, this was by no means obvious to anybody, otherwise we would neither have wasted time in the attempt nor carried it out the way we did.
From this point of view, it is difficult to dispute that the objective was not attained. This does not mean we are done with the tests.
Since the goal was the one reported in the entry summarizing yesterday's shift and repeted here for convenience for everybody who might have overlooked it, we consciously decided to tune things in such a way to have the same modulation indexes (6 and 56) that we had before our shift to make sure we did not have to touch any lock acquisition parameter (but the phase, as duly reported). To make this possible, we played with attenuation levels and left everything else, including the LNFS output levels, untouched.
As discussed at length over and over with all interested parties, and as reported in the documents detailing the RAMS servo performance (see VIR-0627A-21 and VIR-0947A-21), the system originally installed in Nov 2022, and tested yesterday "as is", is designed to have 0.5 W (+27 dBm) at the output of the 6 MHz RF amplifier and 1 W (+30 dBm) at the output of the 56 one.
The RF levels currently used are different, and smaller, than the "nominal" ones.
In these conditions, is not reasonable to expect the RAMS servo to perform as it is designed to do.
If (when?) we want to check that (already done in the past with the same exact unit, by the way, that has been kept installed and off for all this time), it can be of course arranged, making the necessary adjustments to that end though.
In conclusion, saying that "the RAMS is not functioning" at this point is, at the very least, premature.
This plot show the time sequence of the part of the RAMS tests performed on the 2024-01-10
At the end's shift, the RAMS loops were kept closed for the 6MHz and 56MHz modulation .
On the MICH spectrogram one can see that some lines appeared when the 56MHz RAMS was put in operation. The same thing occured one the B4 and B2 12MHz, 50MHz and 112MHz channels:
- plot for B4 and its zoom
- plot for B2 and its zoom
Maybe it could be useful to open the 6MHz and 56MHz RAMS loops and to keep the RAMS noises measurement in operation in order to make more investigations.
