A while ago, Stefan posted the interesting observation that a small increase in the 9MHz modulation depth increased the DARM noise. 45174
I wanted to try to measure the RIN of the sidebands more directly, to compare to the RFAM noise measured by the amplitude stabilization box. I attempted to do this with the interferometer locked at 2W, but I wasn't successful in DC coupling the ISS second loop with only 2W, so the measurements made with the interferometer locked are limited by the intensity noise that is seen by the second loop (the second loop doesn't see the intensity noise on the sidebands due to RFAM since it sees the carrier and sidebands). When locking the OMC on the sidebands at 2W of input power, the OMC length locking loop is also limited by the input intensity noise downconverted around the 4100Hz dither line, so this noise limits the measurement below about 200 Hz and above about 500 Hz.
Instead, I used a single bounce beam, with 20W input power, and the ISS DC coupled with two boosts on. In order to reduce the amount of noise introduced by the OMC length servo, I made a narrower bandpass (which is upstream of the demodulation). I think that we should be able to use this bandpass all the time. For just this test, I also turned off the 1.8 Hz integrator in the OMC length servo, reduced the servo gain by a factor of 4 to get a ugf below 1 Hz, and engaged the ELP10Hz low pass. Locked on the 9MHz sideband, we have 0.34 mA DCPD sum, and locked on the 45MHz we had 0.58 mA. The first attachment shows the reduction in RIN when the control signal is reduced (measured using the 9 MHz sideband, but it was similar for 45MHz). The control signal was reduced enough that we can assume it isn't limiting the measured RIN above 15 Hz or so.
Using the switches on the amplitude stabilization box front panels to enable the excitation, and LSC-EXTRA_AO_3 for the excitations, I was able to excite RF AM for both 9 and 45 MHz separately. Based on the wiki page for the driver (EomDriver), I think that the AC_OUT channel is the out of loop measurement of the amplitude noise on the RF drive. These channels are calibrated into RIN, but they are calibrated for 23dBm for 45MHz (which is what we are using for lock acquisition) and 17dBm (9MHz), while we are using 27dBm for lock acquisition. so the calibrations need to be scaled. For the measurement made with 9 MHz, I scaled the RF9_AC_OUT channel by 2.9RIN/count to get it into units of DCPD rin, and for the 45MHz measurement I used 0.908 RIN/count (this is already calibrated).
The second and third attachments show the RIN measured with and without an excitation for both sidebands. Above 2.5kHz both 9 MHz and 45 MHz sidebands are limited by some intensity noise from the input beam (intensity noise also seen by the ISS second loop). Below 1 kHz, the RIN on both of the sidebands is about a factor of 10 above the ISS out of loop RIN, and well above what would be predicted by the RF AM monitors.
Times:
OMC locked on 9 MHz reference time: 31/12/2018 23:43:51 UTC excitation time: 23:45:20 UTC
OMC locked on 45 MHz reference time: 31/12/2018 23:56:20 UTC excitation time 1/1/2019 00:04:23 UTC
Other notes:
- I noticed in the LSC model that there is a front end calculation of the POP RIN, but it wasn't used because the medm screen was missing, so I added the screen and a low pass.
- I am leaving the narrower bandpass on the the OMC length loop. I doesn't change the DC gain and we should have enough phase margin. Loop measurement attached.
- The interferometer locked for me the first time this morning, with no problems.
- I lost lock when I tried to change the coil driver states to go to low noise, because the dither lines cause them to saturate now. We can move the lines to lower frequencies to avoid that, but for now I have added a path in the guardian around the coil driver switching.
- Since that lockloss, I have had several locklosses when trying to transition DARM from ALS to AS45. I am not sure why this is happening, everything seems OK with the 45MHz EOM driver, and there is a 45MHz signal that is coherent with the ALS signal. It looked like the locklosses were caused by the DARM gain ramoing to 1000, so I tried commenting that out, but we still lost lock.
