Reports until 18:23, Sunday 05 February 2023
H1 AOS
robert.schofield@LIGO.ORG - posted 18:23, Sunday 05 February 2023 - last comment - 06:54, Monday 06 February 2023(67075)
Update on grounding noise studies

Sheila, Robert

Recently, improved grounding of multiple chassis at EX reduced peaks and 20-100 Hz coherence between DARM and a current clamp that monitors current to ground as well as between DARM and electronics racks magnetometers (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=66469).  The improved grounding also reduced the coupling of currents injected onto the Beckhoff ISC Common chassis by a factor of about 5. 

There is a non-zero bias that minimizes ground noise coupling.

I moved on from EX to study EY where I set up a ground current clamp and found coherence in the same 20-100 Hz band. I was surprised to find that coupling of my ground injections at EY, where ETMY normally has a bias of -23 V (we don’t normally use the ETMY ESD), was similar to coupling at EX (see Fig1), where the ETMX bias is about -428 V.  So I scanned the bias at each end station from roughly -400 to +400 while injecting onto the ground. Figure 1 shows that the coupling of my ground injections nearly disappeared at + 115 V (ETMY) and +128 V (ETMX).  We had previously reported that when the bias was reduced by about half, the coupling decreased by a factor of 0.6, which is consistent with zero coupling being at 128V instead of 0V for ETMX.

Figure 1 also shows that the coherence between ground noise and DARM was reduced at EY when the bias was set to the minimum for injection coupling.

Connecting ETMY ESD leads together reduces coupling but not as much as setting the  bias to near the coupling minimum.

We don’t normally use ETMY, so I tried ESD termination schemes that Anamaria has developed at LLO. Connecting the bias "send" and "return" leads together across a resistor, but leaving the quadrants hooked up normally, reduced ground injection coupling by a factor of 0.8. The coupling was also reduced by a factor of about 0.8 when I reconnected the bias but connected the send and return leads of each of the quadrants. When both the bias and the quadrants were terminated in this way, the coupling was reduced from nominal by a factor of 0.2. It may be that this residual 20% is associated with the variation of the potential of the cage and other grounded surfaces in the chamber as ground potential fluctuates,  and not primarily with fluctuation on the ESD traces. 

As a test, I injected onto a cable tray instead of a Beckhoff chassis, with the return from the beam tube in both cases. This injection mainly avoids the Beckhoff-ESD path. Relative to the ground fluctuation measured by the current clamp, the coupling to DARM dropped to about 0.25 of the level for the Beckhoff chassis injection. This is roughly consistent with the 0.2 results from the “termination” method of eliminating the Beckhoff-ESD path in the paragraph above.

Better grounding at EY, as at EX, reduced coupling of ground noise

Figure 3 shows that slightly better grounding of Beckoff chassis at EY reduced coupling to DARM by a factor of about two. A few alligator clips were used for this, but Fil has recently improved the grounding at EY as was done at EX, though I haven’t had a chance to check by how much the coupling has been reduced. I would guess that with good Beckhoff grounding we can almost reach the factor of 4 or 5 reduction made by eliminating the Beckhoff-ESD path.

Ground injections appear on the Electric Field Meter

Figure 4 shows that the ground injections are visible on the EFM. I have measured the coupling of fringing E fields coming in through the view ports, to DARM, finding about 1.6e-16 m/ V/m at about 211 Hz (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=41589) . The predicted amplitude in DARM for coupling from the viewports would be about 1/10 of the observed amplitude of the injection peaks in DARM, so, as expected, the coupling is not from fringing fields coming in through the viewports. But the ground noise does produce fields that, if roughly ten times stronger near the test mass than at the EFM, could roughly account for the noise coupling.

To set the scale:

For my ground injection at 22 Hz,  I get about 0.3mV/m sqrt(Hz) at the EFM,

and, about 12 mV/sqrt(Hz) of fluctuation of the building ground, determined from my measurement of 2 ohms between building “ground” and earth, and the current measured by the current clamp on a grounding cable,

and, about 1.53 mV/sqrt(Hz) of drive fluctuation, according to my understanding of the calibration of  H1:SUS-ETMX_L3_DRIVEALIGN_L_OUT_DQ

Thoughts on the coupling mechanism

While I don’t fully understand the coupling, my best guess is that “ground” potential fluctuations, relative to the charged test mass, modulate forces on the test mass. This would include, and is probably dominated by, local “ground” potential fluctuations in poorly grounded chasis, and, to a lesser extent, building “ground” potential fluctuations associated with current fluctuations across the 2 ohms (measured at EX) between building ground and neutral earth.   The ground fluctuations would couple through Vs, Vb and, to a lesser extent Vref in equation 1 of (https://dcc.ligo.org/LIGO-T1700446).  At the bias voltage that produces minimum coupling of ground noise, the force from the excess charge on the test mass is canceled out by an opposite dipole force associated with the bias-polarized molecules of the test mass. A good test of whether we are still limited by ground noise after improved chassis grounding would be to actuate with the photon calibrator and set the biases for the test masses to the ground noise coupling minima. This would eliminate both the ground noise coupling through the ESD and from grounded structures around the test mass.  I haven’t yet measured the coupling minima for the ITMs. 

Non-image files attached to this report
Comments related to this report
peter.fritschel@LIGO.ORG - 06:54, Monday 06 February 2023 (67232)

Very interesting investigation. Regarding the test suggested at the end of actuating with the photon calibrator (I assume you mean send the DARM feedback to the PCal laser rather than the ETMX ESD): why not alternatively set the biases to the ground noise coupling minima (+115 V for ETMY and +128 V for ETMX), but still use ETMX ESD for DARM feedback? There's certainly still plenty of force range at that bias (+128V vs. -428 V), right? How large are the ETMX ESD quadrant drives in low-noise when the bias is at the nominal -428 V?

Note that if the coupling mechanism is as hypothesized above, then the existence of the ESD, rather than being a noise source, is a tool that allows one to cancel the force on the test mass that comes from fluctuating fields acting on a charged test mass.