That schematic posted by
aviatorhi is close, but not quite the same as the -400 IDGs. But it'll do for a quick explanation.
Starting at the top of the ac generator block, there are the main stator windings. These produce the 115V 400 Hz power. In order to do so, a rotating magnetic field must sweep across them. So on the rotor (just below on the diagram), you see the "generator rotating field" a DC current through this (basically an electromagnet) produces that field. But here's an interesting problem: how do you get DC into the rotating rotor? Back in the old days (and in automotive alternators), they used brushes to feed the dc in from the regulator. But brushes are a maintenance problem, so here they use a neat trick. Just below the aforementioned field, there is a 6 diode rectifier and a three phase exciter winding. This is basically like an inside-out automotive alternator. Inside out, because the output windings and rectifiers spin with the rotor. The exciter field (below that) is logically equivalent to an automotive alternator field. Except, being inside out, the exciter field is stationary, eliminating its set of brushes.
The exciter field here is where this CSD-driven generator differs from Boeing's IDG version. Here, it appears that there are a couple of permanent magnets (PM) and some auxiliary windings. These work together with the voltage regulator to vary the total exciter magnetic field. Which varies the output of the exciter windings, which are rectified into dc and fed to the generator field winding. So by varying the exciter field, you indirectly vary the generator field and finally the generator output voltage.
One difference between the CSD system (above) and the IDG system is the presence of the permanent magnets. These allow the generator to 'bootstrap' itself and reduce the ammount of exciter field current that must be added. The IDG generator has no such magnets in its exciter field. Instead, it has yet another small generator with its own set of permanent magnets on the shaft. This produces a small amount of power (at 1200 Hz IIRC) to the Generator Control Unit. The GCU (which incorporates the voltage regulator) uses this to power up its microprocessor and to provide exciter field current. When the generator field relay opens (it opens the exciter field), the generator effectively goes dead. The residual rotor magnetism is far below that which the CSG PM's produce.
Disconnecting the exciter field is done to stop the generator from producing power (short of mechanically disconnecting it from the engine) should there be a fault in the feeders before the GCB, or in the genertor itself. The CSD design (above) may have been judged to be safe due to the small amount of power the still spinning permanent magnets produce. But as things get bigger (the 747 IDGs are rated at 90 kVA each, about 250 Amps) it is important to remove all excitation in the event of a fault.