Displaying report 1-1 of 1.
Reports until 08:50, Wednesday 20 April 2022
H1 ISC (CAL, ISC)
jeffrey.kissel@LIGO.ORG - posted 08:50, Wednesday 20 April 2022 (62696)
H1 OMC DCPD Transimpedance Rack-side Test Repeated: The Response has Changed, but the Story is Inconsistent and Inconclusive
J. Kissel, B. Ratto

The plot, unfortunately, thickens!

In attempt to verify yesterday's reveal that H1 OMC DCPD signal chain has changed (LHO:62676) we remeasured the transimpedance amplifier in the same way as originally measured on 2022-02-15 (LHO:61810). Then, to show the change, we took the ratio of this new 2022-04-19 measurement against the 2022-02-15 measurement. The results are attached.

The ratio of 2022-02-15 measurement and 2022-04-19 measurement *does* show a change, but it is marked different that the 2022-04-19 / 2022-03-16 full signal chain ratio. We suspect the difference is due to the lack of clear grounding scheme in the rack-side measurement setup.

Check out the first attached .pdf, 20220419_H1_DCPDTransimpedanceAmp_OMCA_DCPD.pdf.

In the first page, we repeat the same plot as in LHO:61810, showing the new 2022-04-19 data for each channel on the left, and the ratio of channels on the right.

In the second and third pages, we show, for each DCPDA and DCPDB channel, respectively,
    (BLUE) The ratio of the two rack-side measurements, and
    (RED) The ratio of the two full-signal chain DAC driven measurements.

While the rack-side measurements do so a change, and at the same level in roughly the same frequency region, it is markedly different that the change seen in the DAC driven measurements.

The second attached .pdf shows the rack-side measurement setup, D1900027-v4_aPLUSO4_D2100630-v3_OMCDCPDTransImpAmp.pdf.

I draw your attention to how the output of the SR785 accessory box is connected to the J7 Test Input of the whitening chassis. Specifically, note that because there's no pin of J7 that is connected to the chassis / board / TIA ground, we're stuck connecting the black shields of the excitation's BNC cables to the metal of the breakout board via clip leads, which we can only *hope* is electrically connected to chassis ground, which in turn is connected to the whitening board GND, which we hope is connected to the in-vacuum transimpedance amplifier's ground. Figure 2 of E2100082 seems to indicate that this is true, but given these wonky results I'm not so sure.

Why is this important? Because the audio-band current from the OMC DCPDs is split between two paths:
    (1) I_GW = which is the "gravitational wave" path, is picked up the transimpedance amplifier, a whitening stage, and converted to differential voltage to be read out by the whitening chassis (and eventually the AA and ADC system), and
    (2) I_DC = which is the "DC" path, which connects to the TIA internal electrical GND via the large L4 inductor.
This is reviewed in detail in G2200551.
Here's my guess: The response of path (1) is heavily influenced by the RLC circuit connected in parallel to GND, created between the RC network input impedance to the transimpedance stage (connected virtually to GND via the transimpedance op-amp), and the R and L of the large L4 inductor. 
those these are connected to the same ground internally to the TIA, if the external Test Excitation is not referenced to that same ground, but instead the the unclear and perhaps evolving ground of the J7 chassis connection, I could imagine that the measurement of that TIA response may be spoiled. 

Even if they *are* connected to the same electrical ground, the insecure clip lead connection to the exposed metal of the breakout board may not have been reproduced identically between the 2022-02-15 and 2022-04-19 measurements.

Further investigation, and perhaps a modified test procedure is definitely warranted.
Non-image files attached to this report
Displaying report 1-1 of 1.