For connecting DC connectors to a cryostat, it's quite common to bring all the DC lines out through a shielded cable and then out to a break-out box. The cryostat's electrical surface is extended through the cable shield out to the breakout box chassis -- this much makes sense. It's common on these breakout boxes to have a pair of switches for each BNC connector -- one that shorts the pin to ground, and the other floats the shield of the BNC connector completely or grounds the shield to an internal ground. A banana connector on the back allows one to short the internal ground to the chassis. The signals broken out do not extend beyond maybe 100 Hz or so, if that. The internal connections go to high impedance loads in the cryostat.
Does this whole scheme defeat the point of the shield? For good EMI shielding, I would expect that you want the BNC shield shorted _all around_ to the chassis, rather than being isolated and then connected through a switch. This arrangement seems to me to add quite a bit of inductance to the path to the chassis. How should I analyze this shielding/grounding arrangement to understand the compromises made?