J. Kissel, P. Fritschel, E. Hall, C. Cahillane

Over the past few months as I've been upgrading my damping loop noise budgets, I've been claiming to have found that the HSTS and HLTS type suspensions P and Y motions are limited by M1 DAC noise below 2.5 Hz -- see the latest versions in LHO:65704 for the HSTS SRM, SR2, and FC1 and LHO:65687 for the HLTS PR3.

However, with the help of Peter today, and Evan and Craig last Friday, I've measured the voltage coming out of top-mass AI chassis for H1 SUS FC1 M1 LF (Friday, 2022-11-10) and much more rigorously H1 SUS PR3 (Today 2022-11-15).

From these results, we can clearly see that 
    - the top mass actuation is NOT limited by DAC noise; it being free of the DAC noise down to 0.1 Hz,
    - the DAC noise models I've been using for both 18- and 20-bit DACs from G1401399 for 18-bit DACs and G1500761 for 20-bit DACs are just wrong; way over estimating the noise below 10 Hz.
 
Further, after probing various configurations of PR3, we're confident that applying a ~7 kHz dither to the 18-bit DACs *would* make the DAC noise much larger, starting to "fill in the troughs" of the DAC request of the DAMPED PR3.
Finally, as a side note, the DAC noise does not appear to depend on whether there's a DC offset.

Let's show some plots. In this main entry, I'll show the "final answer" plots for FC1 M1 LF (driven by a 20-bit DAC) and PR3 M1 LF (driven by an 18-bit DAC). 

In these plots I show:
    - Light grey: the noise floor of the noise measurement tool -- the SR785 -- and its signal input range configuration. 
    - Maroon (dark red): the noise coming out of the DAC with NO requested drive what-so-ever. "It is known" that this is NOT representative of the DAC noise at all times -- since we know DAC noise levels, in general, are quite non-linear -- but it's the first and easiest thing to do, so I show it for reference. Again, we explore different configurations of drive in the comments that convinces us that while not representative during drive request, the DAC noise is *about* this level, and at least sufficiently below the drive request.
    - Gold: the counts of DAC drive requested -- calibrated into output voltage via 20 / 2^18 [V/ct] or 20 / 2^20 [V/ct] -- during the exact same time that the voltage output was *measured*
    - Purple: the measured voltage output during the current "nominal" configuration of the suspension ***
    - Green: proof that the respective 18-bit or 20-bit DAC noise models are quite wrong (though it is some consolation that they do validate my noise budget models, showing that *if* the DAC noise were that high, it *would* be dominated by that noise below 2.5 Hz).

Though you'll believe me more once you see all the DAC request configuration tests we did in the comments -- for now, because the request perfectly matches the voltage measured and one does not sees the "troughs" of the damping loop output request obviously polluted by some flat-ish pink noise: these DOFs are *not* limited by DAC noise.

*** Notes on the "nominal" configuration for each of these suspensions:
    - For FC1, we do not yet apply any alignment offsets or ISC request, so the only requested output is from the damping loops. The seismic system underneath the SUS was performing as best it currently can.
    - For PR3, there are alignment offsets (Y offset 150 urad >> 11566 cts out of the DAC at LF and RT [out of 131k available]) as well as damping loop request. Further -- and I think this may mean this measurement is worth revisiting -- the seismic system underneath PR3 was *not* in its nominal configuration. Because we measured the noise on maintenance day, the corner station sensor correction was OFF, which increases the residual seismic motion in the 0.5 to 5 Hz band -- right where the DAC noise is "close." That being said, I don't anticipate the seismic performance getting so much better with sensor correction that we *would* be dominated by DAC noise.