Mr. Ellsworth confirms this. It does not have to be to any defined ground, just of low effective resistance or impedance so that any current that would have passed through the bearings will now do so through the 'third brush' connection. In either direction... ;-}
In fact there may be conditions under which a defined longer ground path might act as an antenna, and that would likely be counterproductive.


Personally I wouldn't think so; in fact I can't think of many ways you'd get into difficulty even if there were substantial asymmetrical RF picked up in the cabling from the generator to the radio, or induced either in the rotor or field windings. I'm reminded a bit about the old joke of the equipment saving the fast-blow fuse by lunching first: if you have a bidirectional low impedance shunt across the bearings the likelihood of even slight current flow in even typically-lubricated bearings will be minimized. It would be possible to run a higher-dielectric grease in the bearings should any stray current still be observed across them...


The great key here is the use of the voltage-to-voltage converter (assuming it is shielded against emitting RF itself at presumably higher-than-AF switching frequency) which could be built to isolate any spike or impulse noise in the radio from the DC supply coming from the generator. Even if the turbogenerator frame weren't mounted on the substantial electrical 'sink' of the boiler and frame of the locomotive, it is difficult to imagine charging either the rotor or the end bell differentially to high enough voltage to induce preferential sparking if not 'shorted' by the external means.

Many Thanks once again for all the time & brain power put into this discussion. Allow me to recap what my perception of this exchange has been and I'll gladly stand for correction if I have misunderstood anyone.
My conclusion is:
If the wiring on the engine has been properly designed/installed/maintained
If the turbo is in good shape including the insulation of the windings
If the radio is properly installed and antenna system is properly tuned

Then Using a radio is not significantly more likely to cause bearing failure due to arcing than switching other heavy loads such as headlights

And the likeleyhood becomes vanishingly small if a low impedence shunt is applied around the bearings between the rotor shaft and the end bell of the turbo.

It is absolutely true that electricity has been studied for centuries and what we really know of it is still only theory...................mld

Actually, it would be far more accurate to say that electricity has been studied for centuries and most of what we really know of it is not theory, but test, experiment, and empirical rule generation. No one really knows, theoretically or otherwise, what the actual structure of an electron is, and while one can be imaged (e.g. via AFM) the result bears no resemblance to something that can reveal internal structure or composition.

Fortunately electricity, like Angus Sinclair's steam, goes where it wants and not where human circles and arrows say it should go ... and how it does that is as reliable (to truly awesome orders of magnitude) as most anything else we can measure in the Universe. All that is really necessary is to pay attention to what the crackle and hum and measured intensities are telling you ... and keep the smoke contained where it should be.

We even know why some traction headlight shells are at 600V potential but you don't get a lethal shock from them... ;-}