wondering how the brush position is shifted
Imagine an ordinary large DC brush motor with one set of brushes (i.e. one brush arm per pole) on a rocker that can be controlled by a handwheel, but positioned so that they occupy the half of the comm width nearer the bearing. A second rocker, around the outside of the first, is fitted with longer brush arms on a larger PCD so they reach over and past the first set, enabling the brushes run on the half of the comm nearer the winding. The two rockers are geared together in opposite directions and controlled by a handle or servomotor.
They say a picture paints a thousand words. You have the picture already - now I am going to try and walk through how the Schrage motor works in just words, without formulae or proofs. The key to understanding it is to mentally untangle the two separate functions that share a primary winding - the induction motor consisting of the primary and secondary, and the combined frequency convertor and variable transformer consisting of the primary, tertiary and brushgear.
If you start by ignoring the tertiary and brushgear, and instead connect the secondary to a starting resistance, you have a regular wound-rotor induction motor. It's mechanically inside out compared to normal motors - usually you would energise the stator and control the rotor for practical reasons - but theoretically it is the same motor. Normally all we can do is slow the motor by inserting resistance, as in a rotor resistance starter. But the speed of any wound-rotor motor could be controlled by injecting a voltage into the secondary circuit (rotor of a WRIM, stator of a Schrage) of the proper frequency and phase relative to the induced voltage which is at slip frequency. One polarity would slow it below synchronous, the other would speed it above synchronous. But the challenge is to generate this voltage at slip frequency - some conversion mechanism is required that maintains a precise phase relationship to the motor flux wave.
The trick with the Schrage motor is that it functions not only as a motor, but also as its own injection supply frequency convertor. By energising the rotor as the primary instead of the stator, the MMF wave cuts through space not at supply frequency but at slip frequency, therefore the voltage between points on the primary stationary in space is also at slip frequency. The same is true of the tertiary, which is coupled to the primary, so by bringing that out to a commutator a voltage is available from the brushes at slip frequency.
Iif we are to control the speed then that voltage needs to be variable. If we require both sub- and super-synchronous operation it must vary either side of zero, phased w.r.t the flux wave. An external autotransformer would serve the purpose but is unnecessary if the brushes are movable. The angular distance between the two sets controls how much of the tertiary winding is inserted in the secondary circuit, hence how much of its voltage is impressed on the secondary, and also its phase. For differential movement, each phase behaves like a Variac with two brushes, moving apart from the centrepoint, with the secondary slung between them. However the absolute position of the rockers can be moved too, which alters the phase of the voltage relative to the flux wave in the stator. Thus all the requirements are fulfilled for an injection supply capable of raising or lowering the speed of the machine relative to synchronous.
There. Did that make any sense?