Summary
The current of the REFLAIR diode while in lock is about a factor of 10 below where shot noise is equivalent to dark noise.
Details
In order to better understand the out-of loop behavior of the Common Mode loop, I did a lightbulb test on the REFLAIR_A diode. The thermal radiation from the lightbulb acts like a quantum limited light source at 9MHz.
The light was powered by a benchtop DC power supply. I mounted the bulb right in front of the diode and varied the DC light level (as seen by the REFLAIR_LF channel) and recorded the noise floor in the REFLAIR 9MHz demodulated channels. The whitening gain was temporarily changed from 12dB to 42dB for this test to overcome the ADC noise level.
The attached pdf file shows the noise vs current curve, and a fit. EDIT: apparently the LF channel is already calibrated in milliWatts. The 9MHz channel is not calibrated to physical units (though the calibration could be determined from the shot noise). As one can see, the light level we use in lock is about 10 times smaller than where shot noise starts to overcome dark noise.
REFLAIR also has an ND filter stack screwed directly on the diode box. Since we have so little light, maybe we should remove it.
data files
The amount of rf coming out of REFL9 in full lock (at 20+ W) is about −10 dBm, or 70 mV. It should be fine to switch over.
