Flintlock76
No wonder to me, man. Looking at all that (and I have to say it) over-engineering makes me wonder if the designers were playing some kind of expensive practical joke on the C&O.
There is very little actually 'strange' about this, especially when compared with the PRR V1 design that was under consideration about the same time. As I think I've noted, this was a hurry-up Baldwin project from the immediate postwar years intended to 'steal a march' on the (presumed) market for noncondensing turbines that PRR thought it had a commanding lead (and some pending patent protection) on. Apparently much of the development was conducted in a 'hush-hush' atmosphere and this probably contributed to some of the 'unanticipated consequences' issues with the locomotives as built.
The design is remarkably conservative: the pressure used is 'only' 310psi which is ridiculously low for a 6000hp turbine. Cumulative losses in the transmission, and back EMF concerns at speed, eat up a surprising amount of the nominal horsepower, just as they have in every DC turboelectric since the French built the first one in the 1890s.
The first problem Wayne's merry control system runs into is the biggest: Imagine that in your O scale analogy, the 120V supply from the wall is powered from a steam turbine, like a version of a Pyle headlight generator. Whether or not the AC transformer is pulling coupled power, the turbine is spinning the generator core at a speed corresponding to the 60Hz of the wall power ... and passing the corresponding mass flow of steam, and its enthalpy, to exhaust. That's a lot of expensively treated, expensively heated water mass to throw out the stack.
In any case, in this pre-60s era, the large generators are just that: DC generators, and the voltage they make is proportional to speed. The control system used for the turbine governor is probably a version of at least a modified Ward-Leonard scheme (see Erik for better details in comprehensible English) so both the speed and excitation of the generator will be modulated to produce the DC output actually fed to the motors... this may be very similar to load regulation on contemporary Baldwins. The trick here is to minimize 'slip' by getting the turbine rotor to a speed that matches mass flow (thereby getting around the single biggest problem of the direct-drive PRR S2) and then balancing turbine speed and generator excitation as you increase, make transition, etc.
Of course, it's relatively stupid to design a DC-motor STE in the first place: the motors don't like water or coal dust (which is conductive; the two together are not just an open invitation to Flashover City, which ends your locomotive working day, but will build up as deposits in the motor case and then pop loose to a remarkable dead short when you hit a good bump (of which C&O had plenty). Note that C&O didn't have motors on all the effective axles, which was cutting something of an unnecessary corner given all that cast underframe, but was smart enough not to try motoring the truck axles (which N&W was perilously trying only a couple of years later, before they drank the Baldwin Fla-Vor-Aid and built a truck-mounted disaster...)
There were DC engines that used rheostatic control as provided -- the problem with a traction rheostat that big is that it develops a LOT of waste heat on DC, even if it is made of segments with power resistors in between. And that is very expensively created waste heat.
There were a few control schemes that implemented the 'rheostat' as plates in some kind of conductive bath -- the original N&W side-rod things, I think, were controlled that way at slow speed. Of course the idea was to get rid of any variable resistance just as quick as you could to stop throwing away power as heat. Many DC traction controllers have a fixed number of resistances, instead of a relatively stepless resistance like a wirewound pot ... which might not be too happy a thing in a 6000hp locomotive having to start a very heavy consist smoothly.
The later version of the V1 mechanical turbine, reached about the time the Baldwin M1 went into road testing (this is presumably the basis for Loewy's design patent of 1947) had an interesting form of 'direct' speed control: the turbine exerted mechanical torque through a combination of induced magnetic fields, with the result being a combination of torque and rotational speed to a geared final drive with comparatively little electrical loss. This would have been the "9000hp" locomotive referred to in the late-40s PRR book of advanced motive power.
In all likelihood this would have been spooled up to 'appropriate' rpm to avoid slip fairly rapidly, and the Bowes drive then selectively energized to produce high starting torque at appropriate output speed. In this case, control is much as you indicated, with the turbine running at relatively constant speed and any road-speed changes accomplished by modulating the circuits in the Bowes drive, the turbine governor controlling steam flow to keep the turbine in a reasonably constant speed range (corresponding to its best efficiency for given throttle pressure)
