Alexa, Rana, Peter, Evan
Tonight we spent some time looking at REFLAIR_B to see if we could improve DRMI 3f locking. This was motivated by the experience at Livingston, where supposedly too much power on the PD causes RF saturation, which in turn screws up the demod phases of the DRMI length signals.
To test this, we drove PR2 with 300 counts at 131.7 Hz, and SR2 with 300 counts at 183.7 Hz. We then looked at the ASDs of RF135I&Q under the following configurations:
- The first configuration is with no ND filter. Incident power on the PD was 28 mW.
- Rana and Alexa then added an ND1.0 filter in front of the PD, and made sure to direct the reflection from the filter onto a razor blade dump. Incident power on the PD was then 3 mW.
- Rana and Alexa then replaced this with an ND0.5 ND filter. Incident power on the PDwas 8 mW.
The results are summarized in the table below. In each cell, the first number is the height of the peak (in cts/rtHz), and the second number is the pedestal that the peak sits on (also in cts/rtHz), and so the difference gives the net power. The bandwidth for all these ASDs is the same (0.1 Hz), so there should be no bandwidth normalizaton issue.
| No ND | PR2 drive | SR2 drive |
| 135I (cts/rtHz) | 0.26 - 0.30 = 0 | 0.77 - 0.30 = 0.47 |
| 135Q (cts/rtHz) | 0.56 - 0.30 = 0.26 | 0.29 - 0.30 = 0 |
| ND0.5 | PR2 drive | SR2 drive |
| 135I (cts/rtHz) | 0.26 - 0.20 = 0.06 | 0.47 - 0.20 = 0.27 |
| 135Q (cts/rtHz) | 0.62 - 0.20 = 0.32 | 0.22 - 0.20 = 0.02 |
| ND1.0 | PR2 drive | SR2 drive |
| 135I (cts/rtHz) | 0.36 - 0.20 = 0.16 | 0.34 - 0.20 = 0.14 |
| 135Q (cts/rtHz) | 0.29 - 0.20 = 0.09 | 0.60 - 0.20 = 0.30 |
This seems to indicate that the PR2 signal transitions from 135Q to 135I as we lower the power, while the SR2 signal transitions from 135Q to 135I, consistent with the idea that there is some funny business with REFLAIR_B.
We tried locking DRMI on 3f with the ND0.5 filter in place, but found that we could not do it, even after increasing all loop gains by 3 to compensate for the decreased optical power.
The attached spectra perhaps show why. These were taken with DRMI locked on 1f. The dashed spectra are with the ND0.5 filter present, and the solid spectra are with no ND filter. With the ND0.5 filter, RF135I&Q seem to be noise-dominated above 10 Hz or so. Additionally, we note removing the ND0.5 filter does not boost the low-frequency portions of the spectra by 3.4 (=27/8), as we would expect. RF27I appears to be increased by a factor of 5 or so, while RF135Q appears to be increased by a factor of 10 below 1 Hz, and by a factor of 30 at 10 Hz. So there seems to be something strange going on.
