Reports 1-1 of 1 Clear search Modify search
AdV-COM (AdV commissioning (1st part) )
hoak, was, ruggi - 23:00 Thursday 06 April 2017 (37185) Print this report
DC readout with B1s2

Tonight, following the OMC alignment work, Michal locked the first OMC and we transitioned DARM to DC readout using the light reflected off the 2nd OMC.  In this state, the detection PD (B1s2 in this case) is getting about 10x less light than we expect the B1 PDs to receive in nominal low-noise, and we're not getting the benefit of the filtering from the 2nd OMC.  But, by reducing dither lines and tuning the DARM offset, we were able to have B1s2 be free of saturations.  This gave us a much better noise curve compared to last week.  Compared to that curve (with the saturating B1 PDs), the lock of the OMC was more robust and the blend of the B1s2 DC and Audio was free of glitching.

In order to maintain the alignment, we turned off the DIFFp TY loop, and also the PR alignment loops in dark fringe.  Just before transitioning to DC readout we also turned off all of the dither lines on the mirrors (so really only the BS was controlled, with the quadrants).  This isn't a great way to fly, but should be enough to stay locked for a few minutes and measure the noise.

Our best lock, in the lowest noise configuration, was from 19:44:45 UTC to 19:46:40.

First plot attached is a comparison of DARM on RF (purple) to DARM on DC (blue) - there's about three orders of magnitude improvement above 1kHz.  DC readout works!  In this state our strain sensitivity at 1kHz is probably better than 10^-21, we'll see what the calibrators say.

Second plot attached are some spectra and coherences.  The SSFS boost was on during durng this lock, and the DARM noise is perfectly coherent with MICH between 10-100Hz.  Notice the notches in the DARM spectrum at 219 and 237Hz, these are notches in the MICH control filter to stop the MICH->PR feedforward from ringing up the PR violin modes.

We need to fix the MICH->DARM noise before we can hand off to the real detection PDs, the current noise level will saturate the B1 photodiodes.

Images attached to this report
Comments to this report:
mwas - 9:15 Friday 07 April 2017 (37186) Print this report
Finished building the DC read-out model in the noise budget (matching the measurement with one OMC and B1s2 seeing only 8% of the light).

Figure 1, the DARM spectrum calibrated into strain, lowest point at 2e-22! A big caveat is that last week the calibration to strain done in the noise budget, and done by the proper calibration disagree by a factor ~10. We still need to investigate why the disagreement happens and which one is correct.

From 10Hz to 1kHz completely explained by MICH noise and the model and actual coupling are matching very well, after adjusting some of the fudged gains in the model by factors ~2. Above 1kHz the noise floor sits nicely on where we expect to find the quantum and electronic PD noise. The structures above 1kHz look similar to the SSFS projection but a factor 10 higher.

I have put the modeled MICH to DARM transfer function in /users/mwas/ISC/DClock_20170406. There is also there the noise budget result spectra and transfer functions. But this is less practical as it requires the matlab noise budget library in path to open the structures, using for example:
addpath(genpath('/users/mwas/advsw/NoiseBudgets/Externals/SimulinkNb'))
Images attached to this comment
Search Help
×

Warning

×