J. Kissel, M. Pirello, D. Schaetzl, D. Sigg

Over the course of initial testing of the new OMC DCPD whitening chassis (D2200215), whose primary filter board PCB is D2200044, we identified two things we were sad about:
    (1) The channels seemed to have a pretty large self-noise compared to the desired noise level ("below the new low-noise 18-bit ADC, which is ~500 nV/rtHz"), and
    (2) The output voltage had a pretty large DC bias, at around 2.5 V_DIFF.

As such, and through a bit of experimental demonstration, we improved the D2200044 circuit in the S2300003 in the following way:
    - Replaced U2 op amps in whitening board with OP37 (to reduce output voltage offset, noise, and speed of chip)
    - Reduced resistance values of R9 & R10 to 14.3k & 1.6k Ohm respectively, and correspondingly increased C6 cap to 10 uF (to reduce Johnson noise imprinted on total output voltage noise)
    - In doing so, removed TP2 (connected to GND) in order to fit in 10 uF (+/-1%) film cap, and connected GND side of cap into former TP2 position
    - Increased C5 op-amp bandwidth compensating cap to 10 pF (to decrease corresponding R9 & C15 pole from 11 MHz to 1 MHz ?because neither frequency has impact in the audio band, and nothing good can come of leaving signal unfiltered between 1 and 11 MHz)

With this improvement, the output voltage self-noise for the OMC DCPD whitening chassis, with whitening ON, now is around 300 nV/rtHz at ~100 Hz, lower than the expected 18-bit, low-noise ADC.

The attached 2023-01-31_OMCDCPDWhitening_S2300003vsS2300002_OutputVoltageNoise.pdf summarizes the comparison between the S2300002 chassis which *hasn't* been modified, and the *S2300003* chassis.

For posterity, I also add a collection of 4 plots, 2023-01-30_OMCDCPDWhitening_OutputVoltageNoise_IndividualMeasurements.pdf that shows both channels, from both chassis, in both whitening states.

The final .txt attachment just covers my notes on how to run these noise tests.

As you've already seen in LHO:67110, we've installed the improved S2300003 this morning. Refer to that aLOG regarding the improvements in DC bias on the output voltage, but, in short, the DC bias improved from 2.5 V_DIFF to 0.025 V_DIFF.