After much effort to get the correct calibrations, I have updated the plots of the various ASC REFL WFS noises that I have presented in previous alogs (those older plots had some errors that caused some alarm).
Shot Noise:
The four plots labeled "shot noise comparison" (REFL A RF45, REFL B RF45, REFL A RF9, REFL B RF9) show measurements made of the REFL_A/B_RF45/9_I_PIT spectrum at 10W PSL input (8W out of IMC) in the blue trace. This was when the IFO was unlocked, but the PRM was aligned. The orange trace shows a dark measurement taken when the IFO was unlocked and the IMC offline. They are calibrated to W/rtHz. The green trace shows the shot noise estimate calculated from the DC counts of ASC REFL A and B NSUM (~40,000 counts at 10W PSL input).
RF calibration: ADC conversion factor of 6.1e-4 V/ct (16 bit +-20 V), transimpedance of 900 Ohms (RF9) or 650 Ohms (RF45), photodiode responsivity of 0.8 A/W, DC whitening gain of 21 dB (RF9) or 15 dB (RF45)***, and a demod gain of 22 dB.
DC calibration: ADC conversion factor of 6.1e-4 V/ct (16 bit +-20 V), transimpedance of 1 k Ohm, photodiode responsivity of 0.8 A/W
ADC Noise:
The four plots labeled "nadc noise comparison" (REFL A RF45, REFL B RF45, REFL A RF9, REFL B RF9) show measurements made of the REFL_A/B_RF45/9_I_PIT spectrum in counts and a calculated value of the ADC noise with and without the 2 stages of anti-whitening filters applied.
Calculating n_adc: First, I took a measurement of the "dark noise", in other words, the noise present due to electronics and ADC noise, with the whitening and anti-whitening stages on (and the DC whitening gain)
measurement 1 = n_electronics^2 + n_adc^2(AW)^2 [Both whitening and anti-whitening are applied to n_electronics, cancelling out]
Then, I turned off both the whitening and anti-whitening stages
measurement 2 = n_electronics^2 + n_adc^2 [This is the orange trace in the plot]
By subtracting the two, I can get an estimate of n_adc alone. I plotted this estimated value (blue trace), and n_adc*AW (yellow trace), in other words, what this estimated ADC noise looks like when the anti-whitening stages are applied.
Full Lock Spectra:
The four plots labeled "full lock" show the REFL WFS spectrum from May 2021 (REFL A RF45, REFL B RF45, REFL A RF9, REFL B RF9). The shot noise is again calculated from the counts on the REFL A and B NSUM. For the lock it is ~20,000 cts each. These plots are also calibrated to W/rtHz, (see shot noise calibration notes).
There is one additional gain factor in full lock: REFL RF45 has a 1.41 ct/ct gain in lock, REFL RF9 has 1 ct/ct gain.
For simplicity in all these plots, I only chose to show the I pitch trace. However, I took data on I/Q and pit/yaw (they are all very similar as expected).
***Note: as of March when these measurements were started, the DC whitening gain of the RF9 and RF45 channels were 21 and 15 dB, respectively. As of the end of April, they have been lowered to 15 dB and 9 dB respectively, and then compensated with an additional 6 dB digital filter (to avoid saturation due to the high input power). For the calibration though, 21 dB and 15 dB are correct.
