Sheila, Georgia
Yesterday we re-ran the noise budget templates for the following noises:
- LSC (MICH, PRCL, SRCL)
- Laser frequency noise
- Laser intensity noise
- ASC (DHARD, CHARD, MICH)
Only the frequency noise changed dramatically since the old noise budget, it is significantly lower now - presumably due to the increase in CO2_X power.
The updated noise budget, for a time when squeezing is injected, is shown in the first attachment. We are clearly underestimating our shot noise above a few hundred Hz, though between 10 and 300 Hz we estimate the noise well.
To look further into shot noise, today we let the ifo run for 10 mins without injecting squeezing. The noise budget for this time is shown in the second attachment. Without squeezing on our shot noise is correct at all frequencies.
In the first noise budget we assume a constant (frequency independent) squeezing level, calculated from the DCPD BLRMS, which reported 2.9 dB of squeezing. The third attachment compares DARM with squeezing and DARM without squeezing. Squeezed-DARM sits ~2dB below normal shot noise at 200Hz, but only 1dB at 2kHz. This frequency-dependence was attributed to bad frequency noise in the past, but is still present with improved frequency noise. Perhaps we need to look at whether we have a SRCL offset?
We ran the DCPD cross-correlation template during our time without squeezing, to see if this worse sensitivity at high frequency is due to classical noise that is not in the noise budget. The fourth attachment shows correlated noise based on the noise budget (removing the shot noise and dark noise from the unsqueezed noise budget) in yellow, and the correlated noise from the DCPD cross correlation template in purple. We seem to be under-estimated our cross correlated noise by a factor of ~2 at 2kHz. (The discrapancy between DARM from the noise budget and DARM from the cross-correlation template is the difference between the GDS calibration and the CAL-DELTAL_EXTERNAL calibration). This is not enough to explain our mismatched shot noise.
The fifth attachment is the updated ASC noise budget. We are over estimating our noise a bit between 10 and 20 Hz, one explanation could be that the noise budget templates we use to measure coupling functions were run yesterday when the wind was high, while the Measured Noise trace is from the previous lock, which had lower ground motion. We also want to check the PUM DAC noise.
I'm attaching a couple of other figure that were generated when Georgia ran this noise budget that might be useful for reference. The first is the same as Georgia's third plot, but also shows the frqeuency dependence with an attempt at subtracting the correlated noise. This is the correlated noise from the online cross correlation, so the calibration issue shown in Georgia's 4th plot is impacting the subtraction. We did some follow up measurements for this frequency dependence that are shown in 50591 there I did the cross correlation offline which should avoid the calibration issues (or, would have a different calibration issue, in the offline one the subtraction will be limited by the accuracy of the DARM loop model, not by the discrepancies between the online and offline calibration).
The second attachment shows some projections, the ones that involve squeezing should be interpreted with caution.
- The first noise budget in this main alog is overestimating the squeezing level in the quantum noise curve. The noise budget uses the DCPD BLRMS to measure squeezing level and then applies this to the shot noise. It estimated 2.9dB of squeezing which we know is an overestimate. Not ground-breaking but I've attached an updated spectrum.
- In order to play around with detunings better I have copied the gwinc model used in the LLO noise budget to calculate shot noise. Haocun added some nice cross-correlation tools so I copied her version of gwinc to the iscSVN as gwinc_new.m (sorry this is not a good name), and used an updated IFO parameters file called IFOModelH1.m. I made a separate noise budget script for using gwinc, it lives in the usual place:
/ligo/gitcommon/NoiseBudget/simple-noise-budget/restructured_nb/
