Sheila, Georgia, Craig, Danny, Dan

Earlier this afternoon we spent some time with the interferometer locked at 2W RF DARM, and made some measurements of OMC length noise, similar to the measurements described here: 30510  

The attached plot shows the noise measured with the OMC locked on the 45MHz sideband, both on resonance and with a small detuning from resonance.  With the small detuning, the DCPDs see the OMC length noise, the measurements taken when the OMC is on resonance show the noise floor of this measurement of the OMC length noise. You can see that there are a series of lines in the OMC length that weren't present in the OMC length noise measurements from 2016 linked above, and the comb that appears is the same as the comb seen in the PZT3 monitor, 46982. 

The other noticable feature is the wide 180Hz line in the noise measured with the OMC locked on 45 on resonance, which seems very similar to the 180 Hz line which we have been blaming on the modified PZT driver.  When the OMC is locked on the sideband the intensity noise on the sideband around the dither frequency is higher than when we are locked on the carrier, so the dither frequency needs to be increased to get a similar SNR.  We increased the dither amplitude and saw that shoulders on the 180 Hz line were decreased, meaning that they were upconversion of the dither loop sensing noise onto the 180Hz line which seems to be a new feature in the OMC length since O2.  

We also measured the OMC length loop UGF before changing the dither amplitude, see second attachment, and it's UGF is low, around 1 Hz.  In a similar measurement from December the ugf was almost 4Hz, I've put the template I use to measure this in the userapps/lsc/h1/templates directory (this template is not safe for use in full lock, I use it locked on RF).  It seems likely that the reason for the 180 Hz broad peak when we've swapped the driver is that the ugf of the dither loop wasn't compensated for when the dither amplitude changed because of the change to the driver electronics.  

For now we have increased the dither amplitude for locking on the carrier by a factor of 4, which should give us a similar UGF to the one that we had in December.  The 3rd attachment shows that we can reduce the shoulders on the 60Hz and 180 Hz lines by increasing the dither line, (the 90Hz peak is unrelated). The 4th attachment shows that when we continue to increase the dither amplitude we start to see the power lines upconverted as sidebands around the 4100Hz dither in DARM.  

This improvement gives us a few Mpc (about 4Mpc), which is roughly consistent with Jenne's prediction.  It would still be worth spending some time trying to reduce the power lines (and the comb) in the OMC length noise, since there are still shoulders on these peaks, and the OMC length noise spectrum was cleaner in 2016.