Comparing the REFL readback signals from this morning, where there is NO light on the REFL PDs, to a time when we are locked, it seems that the REFL B readout could benefit from some additional whitening. 
In fact, it seems all readbacks would benefit.  
REFL B is not hooked up to a usual ADC chassis, according to Daniel.  It's not so simple to put in additional whitening.
REFL A is hooked up to the usual three stages of (z=10 Hz, p=1 Hz) whitening, and two stages of whitening are currently being used.  Unclear if we would saturate the ADC if we turned the final stage on.
REFL ERR has NO whitening, it's actually just a -60 dB gain, despite the name in the filter module.

Also, both H1:LSC-REFL_A_RF9_I_ERR_DQ and H1:LSC-REFL_B_RF9_I_ERR_DQ are saved at 2048 samples per second.  Would surely be beneficial to bump this up to 16 kHz.