I measured the dark noise on all the RF QPDs at the REFL, POP and AS ports and quantified how much is ADC noise vs electronics dark noise. I changed the whitening gain of the QPDs to determine if the RF QPDs are limited by ADC noise or electronics dark noise. This work is in preparation for moving to the wavefront sensors for the alignment control, so we can understand the noise limits of the wavefront sensors.
Conclusion: all RF QPDs are using a flat frequency independent whitening/anti-whitening scheme. This is sufficient for the time being at the REFL and POP ports. The AS port QPDs are close to being ADC noise limited, but the current whitening gain is high. We should consider moving to the frequency dependent whitening filters, and probably consider 3 stages of whitening/anti-whitening at the AS port. 2 stages will likely be sufficient for REFL and POP. A frequency dependent whitening/antiwhitening scheme will probably be better when we use these QPDs in loop.
I took the IMC offline and shuttered the IR and green shutters for this test. First, I took spectra of all the QPDs at their nominal whitening gain settings. Then, I set the whitening gain to 1 dB and retook the data.
The first set of data will show the electronics dark noise of the PDs and the ADC noise with the whitening/antiwhitening gain applied. The electronics noise experiences both the whitening gain and antiwhitening gain (so ithe whitening effect cancels out). The ADC noise only experiences the antiwhitening gain. This is the "nominal" dark noise level for these PDs. N1 = n_dark + n_adc*(A.W.)
The second set of data will show the electronics dark noise and ADC noise without any whitening gain. N2 = n_dark + n_adc
To calculate the ADC noise level, I can subtract these two measurements and divide out the effect of the anti-whitening gain.
The ADC noise level can then be subtracted from measurement 2 to determine the electronics dark noise level.
When I finished my measurements I made sure to revert all gains properly and reopen the shutters.
I calibrated these measurements from counts to volts with the conversion factor 6.1e-4 V/ct, from a +/-20V, 16 bit ADC.
I made plots of all these results showing the nominal dark noise level, the ADC noise level, and the current ADC noise level with the whitening gain in place. This is a lot of plots, so I am only attaching a select few that I feel are representative.
POP: The first plot is the result of POP QPD1 RF17 Pitch. All POP results were about the same. Currently there is 21 dB of whitening gain applied for all sensors.
REFL: REFL RF17 is using 21 dB gain for both QPD1 and 2, but RF45 has 24 dB gain for QPD1 and 27 dB gain for QPD2. All QPDs are about the same, except QPD2 RF45 which seems to have slightly lower electronics dark noise than the others. I am attaching The QPD2 RF45 plot, and one other representative REFL plot for comparison
AS: All QPDs are using 36 dB gain at the AS port. Again, all QPDs are about the same, except QPD2 RF28. It looks like the ADC noise is slightly elevated compared to the others. I again am attaching the strange plot and one other representative for comparison.
My next goal is to determine the shot noise level.