Wensday morning I did a repeat of the bias reduction tests for the EX ESD described in 66376, this time with a lower bias.  One conlcusion is that we don't see evidence of the ground noise being reduced in DARM with a redcued bias (with the improved grounding 66469).  The unexpected finding is that with the reduced bias and higher DAC drives we have increased noise between 25-35 Hz, which isn't explained by either a calibration error or SRCL/MICH/PRCL coherence. Perhaps this is some DAC crossing noise, or something else (apparently we haven't yet implemented the DAC dither lines for the 20 bit DAC).  

The first screenshot and dtt attachment show the DARM OLG measured with the bias reduced, the DARM gain near the UGF is maintained to within 1% for these three sets of settings. The table below is intended to summarize that when I reduce the bias by half (or quarter) the voltage monitor sees a slightly smaller than linear reduction in the bias voltage, and the actuation strength is again weaker than a linear relationship with bias voltage would predict. 

Next I did a few times of switching between the full bias and quarter bias configuration to make a noise comparison:

• quarter bias: 19:01 UTC (refs 0-6, 37-39)
• full bias: 19:16 UTC (ref 7-13, 34-36)
• quarter bias: 19:35 UTC (ref 14-20, 31-33)
• full bias: 19:57 UTC (ref 21-30 ) 
• quarter bias again: 20:38 UTC (leaving at 1/4 bias for the afternoon). 

The second attachment shows the spectra and coherence comparison.  There are lines that come and go in the different tests, none of 40.625 Hz, 51.9 Hz or 68.125 Hz are consistently coming and going with the bias change. 

I think that the way that I did this should have left the calibration unaffected, since I compensated the actaution strength change with digital gain in drivealign and this kept the gain at the DARM ugf the same within 1%.  However, the third attachment shows the braodband pcal injection for the normal bias (red, -447V) vs the lower bias (-124V in blue).  The ratio of the OAF-CLEAN spectrum at 31Hz is 0.72 (1/4 bias over full bias), while ratio of the DARM/Pcal measurements at 31 Hz is 1.017. 

The 4th attachment shows that the coherence of OAF clean with SRCL is slightly lower in the quarter bias test, indicating that there is an increase in some other noise source 25-35 Hz, not that a slight change in the actuation has disrupted the cleaning making more SRCL noise in DARM.  

The 5th attachment shows the drives to ETMX L3 and L2, with the bias set to 1/4 we have tripled the digital gain, while the DAC request increases by 4 from 8000 cnts to 34,000 counts (4 times higher, well below the 500,000 limit of the 20 bit DAC). The L2 drives are unchanged, confirming that I didn't cause changes to the DARM offloading with the bias change. 

A second test of this (quiet time after PEM injections) after Jenne made some changes to cleaning: 

• 18:38:17 UTC 1-13-2023 UTC (quarter bias, no lines in)
• 18:46 UTC back to full bias