I'm going through my old alog drafts, this is from Feb 2023 but the times here may now be useful information, so I'm posting this incomplete draft from a year ago. 

Last week I did an additional test to follow up on 66843 66751 where excess noise was seen when locking DARM on ETMX with an ESD bias of -125V.   Past checks have shown that this excess noise is not due to a calibration error or due to SRCL or MICH feedforward being mistuned with the reduced bias. 

The first attachment shows that indeed this noise is there repeatably in a way that is similar to what was seem in past tests, although the spectrum did change over the time I was doing this test.  This time I did the measurements in NLN_CAL_MEAS so that ADS and calibration lines were off. 

Although the estimate of the ESD quadratic term in 66843 suggested that the ESD quadratic term shouldn't be large enough to explain this noise, I wanted to repeat the test while driving a line on the ESD strongly enough to see the quadratic coupling to confirm that the model of the quadratic coupling is working well.  The second attachment shows three examples of driving a 15.45 Hz line with the same amplitude (10  counts at LOCK_L_EXC), each with slightly different results. The green cursors show the injected line and it's harmonic at 30.9Hz. The first time that I injected the line, it seemed as though the broad noise from 20-35Hz was reduced (grey trace, compare to yellow and blue), but in the next two tests this was not repeated.  

The third attachment shows what happened when I tripled the amplitude of the injected line, the 15.45 Hz peak is increased by a factor of 3 while the 30.9Hz peak is increased by a factor of 9; this makes sense.