I installed isolated feed-thrus on the outputs of the rms-to-dc converters. The grounds of the BNC connectors carrying the BCM1 signals couldn't be completely isolated from the universe (see below), but it turned out that we had a desirable single point ground anyway. Didn't see any evidence of V/F chatter on the ouput signals, nor power line harmonics for that matter. Hopefully, my messing today won't hurt and may help. Need to remeasure the noise levels when beam comes back.
Started with 0.1 Ohms from rms-to-DC converter outputs to rack ground. Not exactly stellar isolation. Changed to isolated feed-thrus whereupon the resistance to ground increased to 1 Ohms. Disappointing, I was hoping for more. Decided to isolate the entire Euro-crate with nylon screws, etc, achieving 20 kOhm isolation from the rack ground. Great! Turned the power on and rechecked. DVM is too confused to measure the resistance, so checked the voltage. The entire Euro-crate has floated to +6V. Tracked it down to the ground on the MPS BCM power supply that came with the Eurocrate. Held my breath and shorted the power supply ground to real ground thru someone else's ammeter: only 3 mA. So the power supply isn't broken, it just wants you to ground it to something you care about. Consulted with the King. The Qweak ADC inputs are sort of differential, but dragging the ground up to 6V is probably going to shut off their front ends. So regrounded the Eurocrate with a conducting screw. There's about 1 mV of DC offset now between BCM1/2 connector grounds and the rack, presumably due to 1 mA of trickle current across the 1 Ohm resistance on the board. The final score: this afternoon I increased the ground isolation by 1 Ohm at the rate of 0.0001 Ohm/second.